id
string | submitter
string | authors
string | title
string | comments
string | journal-ref
string | doi
string | report-no
string | categories
string | license
string | abstract
string | versions
list | update_date
timestamp[s] | authors_parsed
sequence |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2505.20613 | Ziju Shen | Ziju Shen and Naohao Huang and Fanyi Yang and Yutong Wang and Guoxiong Gao and Tianyi Xu and Jiedong Jiang and Wanyi He and Pu Yang and Mengzhou Sun and Haocheng Ju and Peihao Wu and Bryan Dai and Bin Dong | REAL-Prover: Retrieval Augmented Lean Prover for Mathematical Reasoning | null | null | null | null | cs.CL cs.AI cs.LG cs.LO | http://creativecommons.org/licenses/by/4.0/ | Nowadays, formal theorem provers have made monumental progress on high-school and competition-level mathematics, but few of them generalize to more advanced mathematics. In this paper, we present REAL-Prover, a new open-source stepwise theorem prover for Lean 4 to push this boundary. This prover, based on our fine-tuned large language model (REAL-Prover-v1) and integrated with a retrieval system (Leansearch-PS), notably boosts performance on solving college-level mathematics problems. To train REAL-Prover-v1, we developed HERALD-AF, a data extraction pipeline that converts natural language math problems into formal statements, and a new open-source Lean 4 interactive environment (Jixia-interactive) to facilitate synthesis data collection. In our experiments, our prover using only supervised fine-tune achieves competitive results with a 23.7% success rate (Pass@64) on the ProofNet dataset-comparable to state-of-the-art (SOTA) models. To further evaluate our approach, we introduce FATE-M, a new benchmark focused on algebraic problems, where our prover achieves a SOTA success rate of 56.7% (Pass@64). | [
{
"version": "v1",
"created": "Tue, 27 May 2025 01:26:11 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 03:25:43 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Shen",
"Ziju",
""
],
[
"Huang",
"Naohao",
""
],
[
"Yang",
"Fanyi",
""
],
[
"Wang",
"Yutong",
""
],
[
"Gao",
"Guoxiong",
""
],
[
"Xu",
"Tianyi",
""
],
[
"Jiang",
"Jiedong",
""
],
[
"He",
"Wanyi",
""
],
[
"Yang",
"Pu",
""
],
[
"Sun",
"Mengzhou",
""
],
[
"Ju",
"Haocheng",
""
],
[
"Wu",
"Peihao",
""
],
[
"Dai",
"Bryan",
""
],
[
"Dong",
"Bin",
""
]
] |
2505.20618 | Po Yi Wu | Po-Yi Wu | An Operator-Splitting Scheme for Viscosity Solutions of Constrained Second-Order PDEs | null | null | null | null | math.NA cs.NA | http://creativecommons.org/licenses/by/4.0/ | This work introduces a novel operator-splitting finite element scheme for approximating viscosity solutions of constrained second-order partial differential equations (PDEs). The method is designed for solutions with low regularity in \( C(\overline{\Omega}_T) \cap H^1(\Omega_T) \). By decoupling the PDE evolution from constraint enforcement, the scheme combines a stabilized finite element method with a semi-implicit time-stepping strategy. We prove that the scheme satisfies a discrete comparison principle by showing the discrete operator yields an M-matrix, which guarantees monotonicity and \(L^\infty\)-stability. These properties, along with consistency, ensure convergence to the unique viscosity solution via the Barles-Souganidis framework. For regular solutions, we establish an optimal-order error estimate of \( O(\Delta t + h^2) \). Numerical experiments validate the theory. | [
{
"version": "v1",
"created": "Tue, 27 May 2025 01:52:37 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 02:59:54 GMT"
},
{
"version": "v3",
"created": "Tue, 17 Jun 2025 02:50:43 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Wu",
"Po-Yi",
""
]
] |
2505.20667 | Tianwa Chen | Tianwa Chen, Barbara Weber, Graeme Shanks, Gianluca Demartini, Marta Indulska, Shazia Sadiq | How Do Experts Make Sense of Integrated Process Models? | null | null | 10.1007/978-3-031-94571-7_1 | null | cs.IR cs.HC cs.SE | http://creativecommons.org/licenses/by/4.0/ | A range of integrated modeling approaches have been developed to enable a holistic representation of business process logic together with all relevant business rules. These approaches address inherent problems with separate documentation of business process models and business rules. In this study, we explore how expert process workers make sense of the information provided through such integrated modeling approaches. To do so, we complement verbal protocol analysis with eye-tracking metrics to reveal nuanced user behaviours involved in the main phases of sensemaking, namely information foraging and information processing. By studying expert process workers engaged in tasks based on integrated modeling of business processes and rules, we provide insights that pave the way for a better understanding of sensemaking practices and improved development of business process and business rule integration approaches. Our research underscores the importance of offering personalized support mechanisms that increase the efficacy and efficiency of sensemaking practices for process knowledge workers. | [
{
"version": "v1",
"created": "Tue, 27 May 2025 03:32:28 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 05:39:28 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Chen",
"Tianwa",
""
],
[
"Weber",
"Barbara",
""
],
[
"Shanks",
"Graeme",
""
],
[
"Demartini",
"Gianluca",
""
],
[
"Indulska",
"Marta",
""
],
[
"Sadiq",
"Shazia",
""
]
] |
2505.21121 | Jean-Christophe Pain | Jean-Christophe Pain | Complex pseudo-partition functions in the Configurationally-Resolved Super-Transition-Array approach for radiative opacity | submitted to "High Energy Density Physics" | null | null | null | physics.atom-ph physics.plasm-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A few years ago, Kurzweil and Hazak developed the Configurationally Resolved Super-Transition-Arrays (CRSTA) method for the computation of hot-plasma radiative opacity. Their approach, based on a temporal integration, is an important refinement of the standard Super-Transition-Arrays (STA) approach, which enables one to recover the underlying structure of the STAs, made of unresolved transition arrays. The CRSTA formalism relies on the use of complex pseudo partition functions, depending on the considered one-electron jump. In this article, we find that, despite the imaginary part, the doubly-recursive relation which was introduced in the original STA method to avoid problems due to alternating-sign terms in partition functions, is still applicable, robust, efficient, and exempt of numerical instabilities. This was rather unexpected, in particular because of the occurrence of trigonometric functions, or Chebyshev polynomials, which can be either positive or negative. We also show that, in the complex case, the recursion relation can be presented in a form where the vector of real and imaginary parts at a given iteration is therefore obtained by a sum of the rotated previous ones. | [
{
"version": "v1",
"created": "Tue, 27 May 2025 12:41:45 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 22:00:34 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Pain",
"Jean-Christophe",
""
]
] |
2505.21352 | Chayan Mondal | Chayan Mondal and Sudhanshu Barway | Discovery of a kinematically distinct component in the central region of the collisional ring galaxy AM0644-741 | Accepted for publication in A&A Letters | A&A 698, L25 (2025) | 10.1051/0004-6361/202451984 | null | astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present the discovery of a peculiar central stellar structure in the collisional ring galaxy AM0644-741 using HST imaging and MUSE integral field unit (IFU) data. We identified two S\'ersic components with a S\'ersic index of 1.72 (inner part) and 1.11 (outer part) in the HST F814W band optical image using \textsc{Galfit}. We utilized the MUSE data cube to construct stellar line of sight velocity (V$_{\rm LOS}$), velocity dispersion ($\sigma_{\rm LOS}$), h$_3$ \& h$_4$ velocity moments, and stellar population age maps using the \textsc{GIST} pipeline for further investigating both S\'ersic components, which have a difference of $\sim$ 60 degrees in their position angle. The inner component, with an effective radius $\sim$1 kpc, shows a strong anticorrelation between V$_{\rm LOS}$/$\sigma_{\rm LOS}$ and h$_3$, indicating the presence of a rotating stellar structure. In addition, the inner component also shows a relatively higher velocity dispersion (average values reaching up to $\sim$240 km sec$^{-1}$) along with disky isophotes and stronger Mg~$b$ line strength, which all together highlight a peculiar dynamical state of AM0644-741's central region. Our analysis suggests that the recent encounter has had a smaller impact on the stellar orbits within the inner component. In contrast, it has specifically affected the stellar orbits of the progenitor's outer disk when forming the star-forming ring. The BPT analysis of the unresolved nuclear source shows a LINER-type ionization, hinting at AGN activity in the galaxy. Our study projects the dynamical evolution of collisional systems and provides scope for simulations to explore the central region in greater detail. | [
{
"version": "v1",
"created": "Tue, 27 May 2025 15:46:41 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Mondal",
"Chayan",
""
],
[
"Barway",
"Sudhanshu",
""
]
] |
2505.21652 | Yifan Yin | Yifan Yin, Zhengtao Han, Shivam Aarya, Jianxin Wang, Shuhang Xu, Jiawei Peng, Angtian Wang, Alan Yuille, Tianmin Shu | PartInstruct: Part-level Instruction Following for Fine-grained Robot Manipulation | null | null | null | null | cs.RO cs.AI | http://creativecommons.org/licenses/by/4.0/ | Fine-grained robot manipulation, such as lifting and rotating a bottle to display the label on the cap, requires robust reasoning about object parts and their relationships with intended tasks. Despite recent advances in training general-purpose robot manipulation policies guided by language instructions, there is a notable lack of large-scale datasets for fine-grained manipulation tasks with part-level instructions and diverse 3D object instances annotated with part-level labels. In this work, we introduce PartInstruct, the first large-scale benchmark for training and evaluating fine-grained robot manipulation models using part-level instructions. PartInstruct comprises 513 object instances across 14 categories, each annotated with part-level information, and 1302 fine-grained manipulation tasks organized into 16 task classes. Our training set consists of over 10,000 expert demonstrations synthesized in a 3D simulator, where each demonstration is paired with a high-level task instruction, a chain of base part-based skill instructions, and ground-truth 3D information about the object and its parts. Additionally, we designed a comprehensive test suite to evaluate the generalizability of learned policies across new states, objects, and tasks. We evaluated several state-of-the-art robot manipulation approaches, including end-to-end vision-language policy learning and bi-level planning models for robot manipulation on our benchmark. The experimental results reveal that current models struggle to robustly ground part concepts and predict actions in 3D space, and face challenges when manipulating object parts in long-horizon tasks. | [
{
"version": "v1",
"created": "Tue, 27 May 2025 18:25:42 GMT"
},
{
"version": "v2",
"created": "Sat, 07 Jun 2025 15:34:23 GMT"
},
{
"version": "v3",
"created": "Mon, 16 Jun 2025 19:45:50 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Yin",
"Yifan",
""
],
[
"Han",
"Zhengtao",
""
],
[
"Aarya",
"Shivam",
""
],
[
"Wang",
"Jianxin",
""
],
[
"Xu",
"Shuhang",
""
],
[
"Peng",
"Jiawei",
""
],
[
"Wang",
"Angtian",
""
],
[
"Yuille",
"Alan",
""
],
[
"Shu",
"Tianmin",
""
]
] |
2505.21821 | Takashi Ito | Takashi Ito, Renu Malhotra | Libration of Pluto's argument of perihelion and the role of the major planets | 47 pages, 11 figures, 2 tables. Accepted for publication in Celestial Mechanics and Dynamical Astronomy. This manuscript is the fifth submitted version | null | 10.1007/s10569-025-10249-z | null | astro-ph.EP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Pluto's argument of perihelion is known to librate around $90^\circ$. This libration is related to the secular phenomenon known as the von Zeipel-Lidov-Kozai (vZLK) oscillation. In this work, we make a quantitative assessment of the influence of Neptune's mean motion resonance and of the other giant planets' secular perturbations on the libration of Pluto's argument of perihelion. Here, a parameter $k^2 = (1-e^2) \cos^2 I$ is the key where $e$ is eccentricity and $I$ is inclination. When $k^2$ of a Pluto-like object is larger than a certain critical value, libration of its argument of perihelion would not occur. The secular effect of other disturbing planets (Jupiter, Saturn, Uranus) plays a significant role in determining the critical $k^2$. The non-zero oscillation amplitude of the critical resonant argument also plays a role, although not a dominant one. | [
{
"version": "v1",
"created": "Tue, 27 May 2025 23:07:37 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Ito",
"Takashi",
""
],
[
"Malhotra",
"Renu",
""
]
] |
2505.22743 | Weiyuan Gong | Sitan Chen, Weiyuan Gong, Jonas Haferkamp, Yihui Quek | Information-Computation Gaps in Quantum Learning via Low-Degree Likelihood | 88 pages, 2 figures | null | null | null | quant-ph cs.CC cs.DS cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In a variety of physically relevant settings for learning from quantum data, designing protocols that can computationally efficiently extract information remains largely an art, and there are important cases where we believe this to be impossible, that is, where there is an information-computation gap. While there is a large array of tools in the classical literature for giving evidence for average-case hardness of statistical inference problems, the corresponding tools in the quantum literature are far more limited. One such framework in the classical literature, the low-degree method, makes predictions about hardness of inference problems based on the failure of estimators given by low-degree polynomials. In this work, we extend this framework to the quantum setting.
We establish a general connection between state designs and low-degree hardness. We use this to obtain the first information-computation gaps for learning Gibbs states of random, sparse, non-local Hamiltonians. We also use it to prove hardness for learning random shallow quantum circuit states in a challenging model where states can be measured in adaptively chosen bases. To our knowledge, the ability to model adaptivity within the low-degree framework was open even in classical settings. In addition, we also obtain a low-degree hardness result for quantum error mitigation against strategies with single-qubit measurements.
We define a new quantum generalization of the planted biclique problem and identify the threshold at which this problem becomes computationally hard for protocols that perform local measurements. Interestingly, the complexity landscape for this problem shifts when going from local measurements to more entangled single-copy measurements.
We show average-case hardness for the "standard" variant of Learning Stabilizers with Noise and for agnostically learning product states. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:04:10 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 05:24:14 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Chen",
"Sitan",
""
],
[
"Gong",
"Weiyuan",
""
],
[
"Haferkamp",
"Jonas",
""
],
[
"Quek",
"Yihui",
""
]
] |
2505.22807 | Felipe Areces | Felipe Areces, John C. Duchi | Distribution free M-estimation | 45 pages | null | null | null | math.ST cs.IT cs.LG math.IT stat.TH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The basic question of delineating those statistical problems that are solvable without making any assumptions on the underlying data distribution has long animated statistics and learning theory. This paper characterizes when a convex M-estimation or stochastic optimization problem is solvable in such an assumption-free setting, providing a precise dividing line between solvable and unsolvable problems. The conditions we identify show, perhaps surprisingly, that Lipschitz continuity of the loss being minimized is not necessary for distribution free minimization, and they are also distinct from classical characterizations of learnability in machine learning. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:33:12 GMT"
},
{
"version": "v2",
"created": "Fri, 13 Jun 2025 22:07:28 GMT"
},
{
"version": "v3",
"created": "Tue, 17 Jun 2025 15:41:41 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Areces",
"Felipe",
""
],
[
"Duchi",
"John C.",
""
]
] |
2505.23145 | Jong Chul Ye | Jeongsol Kim, Yeobin Hong, Jong Chul Ye | FlowAlign: Trajectory-Regularized, Inversion-Free Flow-based Image Editing | null | null | null | null | cs.CV cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Recent inversion-free, flow-based image editing methods such as FlowEdit leverages a pre-trained noise-to-image flow model such as Stable Diffusion 3, enabling text-driven manipulation by solving an ordinary differential equation (ODE). While the lack of exact latent inversion is a core advantage of these methods, it often results in unstable editing trajectories and poor source consistency. To address this limitation, we propose FlowAlign, a novel inversion-free flow-based framework for consistent image editing with principled trajectory control. FlowAlign introduces a flow-matching loss as a regularization mechanism to promote smoother and more stable trajectories during the editing process. Notably, the flow-matching loss is shown to explicitly balance semantic alignment with the edit prompt and structural consistency with the source image along the trajectory. Furthermore, FlowAlign naturally supports reverse editing by simply reversing the ODE trajectory, highlighting the reversible and consistent nature of the transformation. Extensive experiments demonstrate that FlowAlign outperforms existing methods in both source preservation and editing controllability. | [
{
"version": "v1",
"created": "Thu, 29 May 2025 06:33:16 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 12:51:50 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Kim",
"Jeongsol",
""
],
[
"Hong",
"Yeobin",
""
],
[
"Ye",
"Jong Chul",
""
]
] |
2505.23182 | Srijith Nair | Srijith Nair, Michael Lin, Peizhong Ju, Amirreza Talebi, Elizabeth Serena Bentley and Jia Liu | FSL-SAGE: Accelerating Federated Split Learning via Smashed Activation Gradient Estimation | Accepted for poster presentation at ICML 2025, 23 pages, 14 figures | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Collaborative training methods like Federated Learning (FL) and Split Learning (SL) enable distributed machine learning without sharing raw data. However, FL assumes clients can train entire models, which is infeasible for large-scale models. In contrast, while SL alleviates the client memory constraint in FL by offloading most training to the server, it increases network latency due to its sequential nature. Other methods address the conundrum by using local loss functions for parallel client-side training to improve efficiency, but they lack server feedback and potentially suffer poor accuracy. We propose FSL-SAGE (Federated Split Learning via Smashed Activation Gradient Estimation), a new federated split learning algorithm that estimates server-side gradient feedback via auxiliary models. These auxiliary models periodically adapt to emulate server behavior on local datasets. We show that FSL-SAGE achieves a convergence rate of $\mathcal{O}(1/\sqrt{T})$, where $T$ is the number of communication rounds. This result matches FedAvg, while significantly reducing communication costs and client memory requirements. Our empirical results also verify that it outperforms existing state-of-the-art FSL methods, offering both communication efficiency and accuracy. | [
{
"version": "v1",
"created": "Thu, 29 May 2025 07:18:59 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 00:36:43 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Nair",
"Srijith",
""
],
[
"Lin",
"Michael",
""
],
[
"Ju",
"Peizhong",
""
],
[
"Talebi",
"Amirreza",
""
],
[
"Bentley",
"Elizabeth Serena",
""
],
[
"Liu",
"Jia",
""
]
] |
2505.23247 | Zonglin Yang | Zonglin Yang, Zhexuan Gu, Houduo Qi, Yancheng Yuan | Accelerating RLHF Training with Reward Variance Increase | null | null | null | null | cs.LG cs.AI math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Reinforcement learning from human feedback (RLHF) is an essential technique for ensuring that large language models (LLMs) are aligned with human values and preferences during the post-training phase. As an effective RLHF approach, group relative policy optimization (GRPO) has demonstrated success in many LLM-based applications. However, efficient GRPO-based RLHF training remains a challenge. Recent studies reveal that a higher reward variance of the initial policy model leads to faster RLHF training. Inspired by this finding, we propose a practical reward adjustment model to accelerate RLHF training by provably increasing the reward variance and preserving the relative preferences and reward expectation. Our reward adjustment method inherently poses a nonconvex optimization problem, which is NP-hard to solve in general. To overcome the computational challenges, we design a novel $O(n \log n)$ algorithm to find a global solution of the nonconvex reward adjustment model by explicitly characterizing the extreme points of the feasible set. As an important application, we naturally integrate this reward adjustment model into the GRPO algorithm, leading to a more efficient GRPO with reward variance increase (GRPOVI) algorithm for RLHF training. As an interesting byproduct, we provide an indirect explanation for the empirical effectiveness of GRPO with rule-based reward for RLHF training, as demonstrated in DeepSeek-R1. Experiment results demonstrate that the GRPOVI algorithm can significantly improve the RLHF training efficiency compared to the original GRPO algorithm. | [
{
"version": "v1",
"created": "Thu, 29 May 2025 08:54:06 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 06:41:40 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Yang",
"Zonglin",
""
],
[
"Gu",
"Zhexuan",
""
],
[
"Qi",
"Houduo",
""
],
[
"Yuan",
"Yancheng",
""
]
] |
2505.23354 | Sai Advaith Maddipatla | Meital Bojan, Sanketh Vedula, Advaith Maddipatla, Nadav Bojan Sellam, Federico Napoli, Paul Schanda, Alex M. Bronstein | Representing local protein environments with atomistic foundation models | null | null | null | null | q-bio.BM cs.AI | http://creativecommons.org/licenses/by/4.0/ | The local structure of a protein strongly impacts its function and interactions with other molecules. Therefore, a concise, informative representation of a local protein environment is essential for modeling and designing proteins and biomolecular interactions. However, these environments' extensive structural and chemical variability makes them challenging to model, and such representations remain under-explored. In this work, we propose a novel representation for a local protein environment derived from the intermediate features of atomistic foundation models (AFMs). We demonstrate that this embedding effectively captures both local structure (e.g., secondary motifs), and chemical features (e.g., amino-acid identity and protonation state). We further show that the AFM-derived representation space exhibits meaningful structure, enabling the construction of data-driven priors over the distribution of biomolecular environments. Finally, in the context of biomolecular NMR spectroscopy, we demonstrate that the proposed representations enable a first-of-its-kind physics-informed chemical shift predictor that achieves state-of-the-art accuracy. Our results demonstrate the surprising effectiveness of atomistic foundation models and their emergent representations for protein modeling beyond traditional molecular simulations. We believe this will open new lines of work in constructing effective functional representations for protein environments. | [
{
"version": "v1",
"created": "Thu, 29 May 2025 11:25:47 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 22:48:20 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Bojan",
"Meital",
""
],
[
"Vedula",
"Sanketh",
""
],
[
"Maddipatla",
"Advaith",
""
],
[
"Sellam",
"Nadav Bojan",
""
],
[
"Napoli",
"Federico",
""
],
[
"Schanda",
"Paul",
""
],
[
"Bronstein",
"Alex M.",
""
]
] |
2505.23471 | Jun Yang | Jun Yang, Cheng-Chi Wang, Bogdan Alexandru Stoica, Kexin Pei | Synthesizing Performance Constraints for Evaluating and Improving Code Efficiency | 30 pages, 3 figures | null | null | null | cs.SE | http://creativecommons.org/licenses/by/4.0/ | Large Language Models (LLMs) have been increasingly used to optimize code efficiency. Evaluating their effectiveness and further suggesting optimization opportunities often rely on high-quality tests to demonstrate the performance bottlenecks presented in the program. However, existing approaches rely on a limited set of hand-curated inputs or LLM-generated uninteresting length-stressing tests, failing to reveal more nuanced optimization opportunities. We present WEDGE, a framework for generating performance-stressing input given the program under test. WEDGE synthesizes explicit performance-characterizing constraints in the form of branch conditions to partition the programs' execution space into performance-specific regions. When integrated with the coverage-guided fuzzer, reaching different regions introduces explicit rewards for test generation to explore inefficient implementations. Our evaluation shows that WEDGE introduces a significant slowdown compared to the tests in CodeContests and those claimed to be optimized by existing approaches. From the utility perspective, integrating our tests substantially improves the existing code optimization approaches that rely on test-driven execution feedback. We release PERFFORGE, the performance tests generated by WEDGE, to benchmark future approaches for efficient code generation at https://github.com/UChiSeclab/perfforge. | [
{
"version": "v1",
"created": "Thu, 29 May 2025 14:26:22 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 16:45:19 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Yang",
"Jun",
""
],
[
"Wang",
"Cheng-Chi",
""
],
[
"Stoica",
"Bogdan Alexandru",
""
],
[
"Pei",
"Kexin",
""
]
] |
2505.23684 | Martin Obaidi | Martin Obaidi, Jakob Droste, Hannah Deters, Marc Herrmann, Raymond Ochsner, Jil Kl\"under, Kurt Schneider | How to Elicit Explainability Requirements? A Comparison of Interviews, Focus Groups, and Surveys | This paper has been accepted at the research track of the 33rd IEEE International Requirements Engineering Conference (RE 2025) | null | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | As software systems grow increasingly complex, explainability has become a crucial non-functional requirement for transparency, user trust, and regulatory compliance. Eliciting explainability requirements is challenging, as different methods capture varying levels of detail and structure. This study examines the efficiency and effectiveness of three commonly used elicitation methods - focus groups, interviews, and online surveys - while also assessing the role of taxonomy usage in structuring and improving the elicitation process. We conducted a case study at a large German IT consulting company, utilizing a web-based personnel management software. A total of two focus groups, 18 interviews, and an online survey with 188 participants were analyzed. The results show that interviews were the most efficient, capturing the highest number of distinct needs per participant per time spent. Surveys collected the most explanation needs overall but had high redundancy. Delayed taxonomy introduction resulted in a greater number and diversity of needs, suggesting that a two-phase approach is beneficial. Based on our findings, we recommend a hybrid approach combining surveys and interviews to balance efficiency and coverage. Future research should explore how automation can support elicitation and how taxonomies can be better integrated into different methods. | [
{
"version": "v1",
"created": "Thu, 29 May 2025 17:23:14 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 11:01:21 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Obaidi",
"Martin",
""
],
[
"Droste",
"Jakob",
""
],
[
"Deters",
"Hannah",
""
],
[
"Herrmann",
"Marc",
""
],
[
"Ochsner",
"Raymond",
""
],
[
"Klünder",
"Jil",
""
],
[
"Schneider",
"Kurt",
""
]
] |
2505.23739 | Connar Rowan Dr | Connar Rowan and Henry Whitehead and Gaia Fabj and Philip Kirkeberg and Martin E. Pessah and Bence Kocsis | Hydrodynamic simulations of black hole evolution in AGN discs I: orbital alignment of highly inclined satellites | 13 pages, 9 figures, submitted to MNRAS | null | null | null | astro-ph.HE astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | The frequency of compact object interactions in AGN discs is naturally tied to the number of objects embedded within it. We investigate the evolution of black holes in the nuclear stellar cluster on inclined orbits to the AGN disc by performing adiabatic hydrodynamical simulations of isolated black hole disc crossings over a range of disc densities and inclinations $i\in[2^\circ,15^\circ]$. We find radiation dominates the pressure in the wake that forms around the BH across the full inclination and disc density range. We identify no well defined steady state wake morphology due to the thin geometry of the disc and the vertical exponential density drop off, where the wake morphology depends on the vertical depth of the transit within the disc. The inclination damping $\Delta i$ relative the pre-transit inclination behaves as a power law in $\sin(i)$ and the ambient Hill mass $m_\text{H,0}$ as $\Delta i/i \propto m_{\rm H,0}^{0.4} \sin(i)^{-2.7}$. The drag on the BH is dominated by the gravity of the wake for the majority of our inclination range until accretion effects become comparable at $\sin(i)\gtrsim30H_0/R_0$, where $H_0/R_0$ is the disc aspect ratio. At low inclinations ($\sin(i)\lesssim3H_0/R_0$) the wake morphology becomes more spherical, leading to a regime change in the inclination damping behaviour. Our results suggest that the inclination damping timescale is shorter than expected from only episodic Bondi-Hoyle-Lyttelton accretion events during each transit, implying inclined objects may captured by the AGN disc earlier in its lifetime than previously thought. | [
{
"version": "v1",
"created": "Thu, 29 May 2025 17:58:03 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 10:20:41 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Rowan",
"Connar",
""
],
[
"Whitehead",
"Henry",
""
],
[
"Fabj",
"Gaia",
""
],
[
"Kirkeberg",
"Philip",
""
],
[
"Pessah",
"Martin E.",
""
],
[
"Kocsis",
"Bence",
""
]
] |
2505.23873 | Haolang Lu | Hongrui Peng, Haolang Lu, Yuanlong Yu, Weiye Fu, Kun Wang, and Guoshun Nan | KGMark: A Diffusion Watermark for Knowledge Graphs | Accepted by ICML2025 | null | null | null | cs.CR cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Knowledge graphs (KGs) are ubiquitous in numerous real-world applications, and watermarking facilitates protecting intellectual property and preventing potential harm from AI-generated content. Existing watermarking methods mainly focus on static plain text or image data, while they can hardly be applied to dynamic graphs due to spatial and temporal variations of structured data. This motivates us to propose KGMARK, the first graph watermarking framework that aims to generate robust, detectable, and transparent diffusion fingerprints for dynamic KG data. Specifically, we propose a novel clustering-based alignment method to adapt the watermark to spatial variations. Meanwhile, we present a redundant embedding strategy to harden the diffusion watermark against various attacks, facilitating the robustness of the watermark to the temporal variations. Additionally, we introduce a novel learnable mask matrix to improve the transparency of diffusion fingerprints. By doing so, our KGMARK properly tackles the variation challenges of structured data. Experiments on various public benchmarks show the effectiveness of our proposed KGMARK. Our code is available at https://github.com/phrara/kgmark. | [
{
"version": "v1",
"created": "Thu, 29 May 2025 12:28:05 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 06:16:20 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Peng",
"Hongrui",
""
],
[
"Lu",
"Haolang",
""
],
[
"Yu",
"Yuanlong",
""
],
[
"Fu",
"Weiye",
""
],
[
"Wang",
"Kun",
""
],
[
"Nan",
"Guoshun",
""
]
] |
2505.23999 | Kuldeep Saha | Shital Lawande and Kuldeep Saha | A topologically extendible mapping class that is not smoothly extendible | The statements of the main theorems were found to be incorrect. The Dehn twist along the meridinal curve is not extendible even topologically | null | null | null | math.GT | http://creativecommons.org/licenses/by/4.0/ | We give an example of a smooth characteristic embedding of a torus in $\s^2 \times \s^2 \# \s^1 \times \s^3$ such that there exists no diffeomorphism of the ambient $4$-manifold that induces the Dehn twist along a meridian of the torus, but there exists a homeomorphism of the ambient $4$-manifold, isotopic to identity, that induces the Dehn twist. As an application of our methods, we provide examples of two proper smooth embeddings of an annulus in $\s^2 \times \s^2 \# \s^1 \times \s^3 \setminus int(\D^4)$ which are topologically isotopic, but not smoothly isotopic (relative to boundary). | [
{
"version": "v1",
"created": "Thu, 29 May 2025 20:48:18 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 08:57:27 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Lawande",
"Shital",
""
],
[
"Saha",
"Kuldeep",
""
]
] |
2505.24120 | Ai Jian | Ai Jian, Weijie Qiu, Xiaokun Wang, Peiyu Wang, Yunzhuo Hao, Jiangbo Pei, Yichen Wei, Yi Peng, Xuchen Song | CSVQA: A Chinese Multimodal Benchmark for Evaluating STEM Reasoning Capabilities of VLMs | 36 pages | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Vision-Language Models (VLMs) have demonstrated remarkable progress in multimodal understanding, yet their capabilities for scientific reasoning remain inadequately assessed. Current multimodal benchmarks predominantly evaluate generic image comprehension or text-driven reasoning, lacking authentic scientific contexts that require domain-specific knowledge integration with visual evidence analysis. To fill this gap, we present CSVQA, a diagnostic multimodal benchmark specifically designed for evaluating scientific reasoning through domain-grounded visual question answering. Our benchmark features 1,378 carefully constructed question-answer pairs spanning diverse STEM disciplines, each demanding domain knowledge, integration of visual evidence, and higher-order reasoning. Compared to prior multimodal benchmarks, CSVQA places greater emphasis on real-world scientific content and complex reasoning. We additionally propose a rigorous evaluation protocol to systematically assess whether model predictions are substantiated by valid intermediate reasoning steps based on curated explanations. Our comprehensive evaluation of 15 VLMs on this benchmark reveals notable performance disparities, as even the top-ranked proprietary model attains only 49.6% accuracy. This empirical evidence underscores the pressing need for advancing scientific reasoning capabilities in VLMs. Our CSVQA is released at https://huggingface.co/datasets/Skywork/CSVQA | [
{
"version": "v1",
"created": "Fri, 30 May 2025 01:34:25 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 01:56:01 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Jian",
"Ai",
""
],
[
"Qiu",
"Weijie",
""
],
[
"Wang",
"Xiaokun",
""
],
[
"Wang",
"Peiyu",
""
],
[
"Hao",
"Yunzhuo",
""
],
[
"Pei",
"Jiangbo",
""
],
[
"Wei",
"Yichen",
""
],
[
"Peng",
"Yi",
""
],
[
"Song",
"Xuchen",
""
]
] |
2505.24164 | Shilin Xu | Shilin Xu and Yanwei Li and Rui Yang and Tao Zhang and Yueyi Sun and Wei Chow and Linfeng Li and Hang Song and Qi Xu and Yunhai Tong and Xiangtai Li and Hao Fei | Mixed-R1: Unified Reward Perspective For Reasoning Capability in Multimodal Large Language Models | null | null | null | arxiv:2505.24164 | cs.CL cs.CV | http://creativecommons.org/licenses/by/4.0/ | Recent works on large language models (LLMs) have successfully demonstrated the emergence of reasoning capabilities via reinforcement learning (RL). Although recent efforts leverage group relative policy optimization (GRPO) for MLLMs post-training, they constantly explore one specific aspect, such as grounding tasks, math problems, or chart analysis. There are no works that can leverage multi-source MLLM tasks for stable reinforcement learning. In this work, we present a unified perspective to solve this problem. We present Mixed-R1, a unified yet straightforward framework that contains a mixed reward function design (Mixed-Reward) and a mixed post-training dataset (Mixed-45K). We first design a data engine to select high-quality examples to build the Mixed-45K post-training dataset. Then, we present a Mixed-Reward design, which contains various reward functions for various MLLM tasks. In particular, it has four different reward functions: matching reward for binary answer or multiple-choice problems, chart reward for chart-aware datasets, IoU reward for grounding problems, and open-ended reward for long-form text responses such as caption datasets. To handle the various long-form text content, we propose a new open-ended reward named Bidirectional Max-Average Similarity (BMAS) by leveraging tokenizer embedding matching between the generated response and the ground truth. Extensive experiments show the effectiveness of our proposed method on various MLLMs, including Qwen2.5-VL and Intern-VL on various sizes. Our dataset and model are available at https://github.com/xushilin1/mixed-r1. | [
{
"version": "v1",
"created": "Fri, 30 May 2025 03:11:46 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Xu",
"Shilin",
""
],
[
"Li",
"Yanwei",
""
],
[
"Yang",
"Rui",
""
],
[
"Zhang",
"Tao",
""
],
[
"Sun",
"Yueyi",
""
],
[
"Chow",
"Wei",
""
],
[
"Li",
"Linfeng",
""
],
[
"Song",
"Hang",
""
],
[
"Xu",
"Qi",
""
],
[
"Tong",
"Yunhai",
""
],
[
"Li",
"Xiangtai",
""
],
[
"Fei",
"Hao",
""
]
] |
2506.00493 | Tim Bedding | Hans Kjeldsen, Timothy R. Bedding, Yaguang Li, Frank Grundahl, Mads Fredslund Andersen, Duncan J. Wright, Jack Soutter, Robert Wittenmyer, Claudia Reyes, Dennis Stello, Courtney Crawford, Yixiao Zhou, Mathieu Clerte, Pere L. Palle, Sergio Simon-Diaz, Joergen Christensen-Dalsgaard, Rasmus Handberg, Hasse Hansen, Paul Heeren, Jens Jessen-Hansen, Mikkel N. Lund, Mia S. Lundkvist, Karsten Brogaard, Rene Tronsgaard, Jonatan Rudrasingam, Luca Casagrande, Jonathan Horner, Daniel Huber, John Lattanzio, Sarah L. Martell, Simon J. Murphy | Asteroseismology of the G8 subgiant beta Aquilae with SONG-Tenerife, SONG-Australia and TESS | accepted by A&A | null | null | null | astro-ph.SR | http://creativecommons.org/licenses/by/4.0/ | We present time-series radial velocities of the G8 subgiant star beta Aql obtained in 2022 and 2023 using SONG-Tenerife and, for the first time, SONG-Australia. We also analyse a sector of TESS photometry that overlapped with the 2022 SONG data. The resulting power spectrum clearly shows solar-like oscillations centred at 430 muHz. The TESS light curve shows the oscillations at lower signal-to-noise, reflecting the fact that photometric measurements are much more affected by the granulation background than are radial velocities. The simultaneous observations in velocity and photometry represent the best such measurements for any star apart from the Sun. They allowed us to measure the ratio between the bolometric photometric amplitude and the velocity amplitude to be 26.6 +/- 3.1 ppm/(m/s). We measured this ratio for the Sun from published SOHO data to be 19.5 +/- 0.7 ppm/(m/s) and, after accounting for the difference in effective temperatures of and the Sun, these values align with expectations. In both the Sun and beta Aql, the photometry-to-velocity ratio appears to be a function of frequency. We also measured the phase shift of the oscillations in beta Aql between SONG and TESS to be -113 +/- 7 deg, which agrees with the value for the Sun and also with a 3-D simulation of a star with similar properties to beta Aql. Importantly for exoplanet searches, we argue that simultaneous photometry can be used to predict the contribution of oscillations to radial velocities. We measured frequencies for 22 oscillation modes in beta Aql and carried out asteroseismic modelling, yielding an excellent fit to the frequencies. We derived accurate values for the mass and age, and were able to place quite strong constraints on the mixing-length parameter. Finally, we show that the oscillation properties of beta Aql are very similar to stars in the open cluster M67. | [
{
"version": "v1",
"created": "Sat, 31 May 2025 10:01:21 GMT"
},
{
"version": "v2",
"created": "Tue, 03 Jun 2025 23:17:31 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 04:19:59 GMT"
},
{
"version": "v4",
"created": "Mon, 16 Jun 2025 23:58:33 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Kjeldsen",
"Hans",
""
],
[
"Bedding",
"Timothy R.",
""
],
[
"Li",
"Yaguang",
""
],
[
"Grundahl",
"Frank",
""
],
[
"Andersen",
"Mads Fredslund",
""
],
[
"Wright",
"Duncan J.",
""
],
[
"Soutter",
"Jack",
""
],
[
"Wittenmyer",
"Robert",
""
],
[
"Reyes",
"Claudia",
""
],
[
"Stello",
"Dennis",
""
],
[
"Crawford",
"Courtney",
""
],
[
"Zhou",
"Yixiao",
""
],
[
"Clerte",
"Mathieu",
""
],
[
"Palle",
"Pere L.",
""
],
[
"Simon-Diaz",
"Sergio",
""
],
[
"Christensen-Dalsgaard",
"Joergen",
""
],
[
"Handberg",
"Rasmus",
""
],
[
"Hansen",
"Hasse",
""
],
[
"Heeren",
"Paul",
""
],
[
"Jessen-Hansen",
"Jens",
""
],
[
"Lund",
"Mikkel N.",
""
],
[
"Lundkvist",
"Mia S.",
""
],
[
"Brogaard",
"Karsten",
""
],
[
"Tronsgaard",
"Rene",
""
],
[
"Rudrasingam",
"Jonatan",
""
],
[
"Casagrande",
"Luca",
""
],
[
"Horner",
"Jonathan",
""
],
[
"Huber",
"Daniel",
""
],
[
"Lattanzio",
"John",
""
],
[
"Martell",
"Sarah L.",
""
],
[
"Murphy",
"Simon J.",
""
]
] |
2506.00555 | Peng Xia | Peng Xia, Jinglu Wang, Yibo Peng, Kaide Zeng, Xian Wu, Xiangru Tang, Hongtu Zhu, Yun Li, Shujie Liu, Yan Lu, Huaxiu Yao | MMedAgent-RL: Optimizing Multi-Agent Collaboration for Multimodal Medical Reasoning | null | null | null | null | cs.LG cs.AI cs.CL cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Medical Large Vision-Language Models (Med-LVLMs) have shown strong potential in multimodal diagnostic tasks. However, existing single-agent models struggle to generalize across diverse medical specialties, limiting their performance. Recent efforts introduce multi-agent collaboration frameworks inspired by clinical workflows, where general practitioners (GPs) and specialists interact in a fixed sequence. Despite improvements, these static pipelines lack flexibility and adaptability in reasoning. To address this, we propose MMedAgent-RL, a reinforcement learning (RL)-based multi-agent framework that enables dynamic, optimized collaboration among medical agents. Specifically, we train two GP agents based on Qwen2.5-VL via RL: the triage doctor learns to assign patients to appropriate specialties, while the attending physician integrates the judgments from multi-specialists and its own knowledge to make final decisions. To address the inconsistency in specialist outputs, we introduce a curriculum learning (CL)-guided RL strategy that progressively teaches the attending physician to balance between imitating specialists and correcting their mistakes. Experiments on five medical VQA benchmarks demonstrate that MMedAgent-RL not only outperforms both open-source and proprietary Med-LVLMs, but also exhibits human-like reasoning patterns. Notably, it achieves an average performance gain of 20.7% over supervised fine-tuning baselines. | [
{
"version": "v1",
"created": "Sat, 31 May 2025 13:22:55 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 03:59:45 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Xia",
"Peng",
""
],
[
"Wang",
"Jinglu",
""
],
[
"Peng",
"Yibo",
""
],
[
"Zeng",
"Kaide",
""
],
[
"Wu",
"Xian",
""
],
[
"Tang",
"Xiangru",
""
],
[
"Zhu",
"Hongtu",
""
],
[
"Li",
"Yun",
""
],
[
"Liu",
"Shujie",
""
],
[
"Lu",
"Yan",
""
],
[
"Yao",
"Huaxiu",
""
]
] |
2506.00854 | Chi-Sheng Chen | Jacky Tai-Yu Lu and Jung Chiang and Chi-Sheng Chen and Anna Nai-Yun Tung and Hsiang Wei Hu and Yuan Chiao Cheng | EEG2TEXT-CN: An Exploratory Study of Open-Vocabulary Chinese Text-EEG Alignment via Large Language Model and Contrastive Learning on ChineseEEG | null | null | null | null | cs.CL cs.AI cs.LG cs.MM q-bio.NC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose EEG2TEXT-CN, which, to the best of our knowledge, represents one of the earliest open-vocabulary EEG-to-text generation frameworks tailored for Chinese. Built on a biologically grounded EEG encoder (NICE-EEG) and a compact pretrained language model (MiniLM), our architecture aligns multichannel brain signals with natural language representations via masked pretraining and contrastive learning. Using a subset of the ChineseEEG dataset, where each sentence contains approximately ten Chinese characters aligned with 128-channel EEG recorded at 256 Hz, we segment EEG into per-character embeddings and predict full sentences in a zero-shot setting. The decoder is trained with teacher forcing and padding masks to accommodate variable-length sequences. Evaluation on over 1,500 training-validation sentences and 300 held-out test samples shows promising lexical alignment, with a best BLEU-1 score of 6.38\%. While syntactic fluency remains a challenge, our findings demonstrate the feasibility of non-phonetic, cross-modal language decoding from EEG. This work opens a new direction in multilingual brain-to-text research and lays the foundation for future cognitive-language interfaces in Chinese. | [
{
"version": "v1",
"created": "Sun, 01 Jun 2025 06:26:32 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 04:58:31 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Lu",
"Jacky Tai-Yu",
""
],
[
"Chiang",
"Jung",
""
],
[
"Chen",
"Chi-Sheng",
""
],
[
"Tung",
"Anna Nai-Yun",
""
],
[
"Hu",
"Hsiang Wei",
""
],
[
"Cheng",
"Yuan Chiao",
""
]
] |
2506.01391 | Zhong Zhang | Zhong Zhang, Yaxi Lu, Yikun Fu, Yupeng Huo, Shenzhi Yang, Yesai Wu, Han Si, Xin Cong, Haotian Chen, Yankai Lin, Jie Xie, Wei Zhou, Wang Xu, Yuanheng Zhang, Zhou Su, Zhongwu Zhai, Xiaoming Liu, Yudong Mei, Jianming Xu, Hongyan Tian, Chongyi Wang, Chi Chen, Yuan Yao, Zhiyuan Liu, Maosong Sun | AgentCPM-GUI: Building Mobile-Use Agents with Reinforcement Fine-Tuning | Updated results in Table 2 and Table 3; The project is available at https://github.com/OpenBMB/AgentCPM-GUI | null | null | null | cs.AI cs.CL cs.CV cs.HC | http://creativecommons.org/licenses/by/4.0/ | The recent progress of large language model agents has opened new possibilities for automating tasks through graphical user interfaces (GUIs), especially in mobile environments where intelligent interaction can greatly enhance usability. However, practical deployment of such agents remains constrained by several key challenges. Existing training data is often noisy and lack semantic diversity, which hinders the learning of precise grounding and planning. Models trained purely by imitation tend to overfit to seen interface patterns and fail to generalize in unfamiliar scenarios. Moreover, most prior work focuses on English interfaces while overlooks the growing diversity of non-English applications such as those in the Chinese mobile ecosystem. In this work, we present AgentCPM-GUI, an 8B-parameter GUI agent built for robust and efficient on-device GUI interaction. Our training pipeline includes grounding-aware pre-training to enhance perception, supervised fine-tuning on high-quality Chinese and English trajectories to imitate human-like actions, and reinforcement fine-tuning with GRPO to improve reasoning capability. We also introduce a compact action space that reduces output length and supports low-latency execution on mobile devices. AgentCPM-GUI achieves state-of-the-art performance on five public benchmarks and a new Chinese GUI benchmark called CAGUI, reaching $96.9\%$ Type-Match and $91.3\%$ Exact-Match. To facilitate reproducibility and further research, we publicly release all code, model checkpoint, and evaluation data. | [
{
"version": "v1",
"created": "Mon, 02 Jun 2025 07:30:29 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 02:57:04 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Zhang",
"Zhong",
""
],
[
"Lu",
"Yaxi",
""
],
[
"Fu",
"Yikun",
""
],
[
"Huo",
"Yupeng",
""
],
[
"Yang",
"Shenzhi",
""
],
[
"Wu",
"Yesai",
""
],
[
"Si",
"Han",
""
],
[
"Cong",
"Xin",
""
],
[
"Chen",
"Haotian",
""
],
[
"Lin",
"Yankai",
""
],
[
"Xie",
"Jie",
""
],
[
"Zhou",
"Wei",
""
],
[
"Xu",
"Wang",
""
],
[
"Zhang",
"Yuanheng",
""
],
[
"Su",
"Zhou",
""
],
[
"Zhai",
"Zhongwu",
""
],
[
"Liu",
"Xiaoming",
""
],
[
"Mei",
"Yudong",
""
],
[
"Xu",
"Jianming",
""
],
[
"Tian",
"Hongyan",
""
],
[
"Wang",
"Chongyi",
""
],
[
"Chen",
"Chi",
""
],
[
"Yao",
"Yuan",
""
],
[
"Liu",
"Zhiyuan",
""
],
[
"Sun",
"Maosong",
""
]
] |
2506.01413 | Yulei Qin | Yulei Qin, Gang Li, Zongyi Li, Zihan Xu, Yuchen Shi, Zhekai Lin, Xiao Cui, Ke Li, Xing Sun | Incentivizing Reasoning for Advanced Instruction-Following of Large Language Models | 13 pages of main body, 3 tables, 5 figures, 45 pages of appendix | null | null | null | cs.CV cs.AI cs.CL cs.LG | http://creativecommons.org/licenses/by/4.0/ | Existing large language models (LLMs) face challenges of following complex instructions, especially when multiple constraints are present and organized in paralleling, chaining, and branching structures. One intuitive solution, namely chain-of-thought (CoT), is expected to universally improve capabilities of LLMs. However, we find that the vanilla CoT exerts a negative impact on performance due to its superficial reasoning pattern of simply paraphrasing the instructions. It fails to peel back the compositions of constraints for identifying their relationship across hierarchies of types and dimensions. To this end, we propose a systematic method to boost LLMs in dealing with complex instructions via incentivizing reasoning for test-time compute scaling. First, we stem from the decomposition of complex instructions under existing taxonomies and propose a reproducible data acquisition method. Second, we exploit reinforcement learning (RL) with verifiable rule-centric reward signals to cultivate reasoning specifically for instruction following. We address the shallow, non-essential nature of reasoning under complex instructions via sample-wise contrast for superior CoT enforcement. We also exploit behavior cloning of experts to facilitate steady distribution shift from fast-thinking LLMs to skillful reasoners. Extensive evaluations on seven comprehensive benchmarks confirm the validity of the proposed method, where a 1.5B LLM achieves 11.74% gains with performance comparable to a 8B LLM. Codes and data will be available later (under review).
Keywords: reinforcement learning with verifiable rewards (RLVR), instruction following, complex instructions | [
{
"version": "v1",
"created": "Mon, 02 Jun 2025 08:11:44 GMT"
},
{
"version": "v2",
"created": "Thu, 12 Jun 2025 13:57:57 GMT"
},
{
"version": "v3",
"created": "Mon, 16 Jun 2025 07:40:34 GMT"
},
{
"version": "v4",
"created": "Tue, 17 Jun 2025 08:52:06 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Qin",
"Yulei",
""
],
[
"Li",
"Gang",
""
],
[
"Li",
"Zongyi",
""
],
[
"Xu",
"Zihan",
""
],
[
"Shi",
"Yuchen",
""
],
[
"Lin",
"Zhekai",
""
],
[
"Cui",
"Xiao",
""
],
[
"Li",
"Ke",
""
],
[
"Sun",
"Xing",
""
]
] |
2506.01565 | Li Zhou | Li Zhou, Lutong Yu, Dongchu Xie, Shaohuan Cheng, Wenyan Li, Haizhou Li | Hanfu-Bench: A Multimodal Benchmark on Cross-Temporal Cultural Understanding and Transcreation | cultural analysis, cultural visual understanding, cultural image transcreation (update dataset license) | null | null | null | cs.CL cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Culture is a rich and dynamic domain that evolves across both geography and time. However, existing studies on cultural understanding with vision-language models (VLMs) primarily emphasize geographic diversity, often overlooking the critical temporal dimensions. To bridge this gap, we introduce Hanfu-Bench, a novel, expert-curated multimodal dataset. Hanfu, a traditional garment spanning ancient Chinese dynasties, serves as a representative cultural heritage that reflects the profound temporal aspects of Chinese culture while remaining highly popular in Chinese contemporary society. Hanfu-Bench comprises two core tasks: cultural visual understanding and cultural image transcreation.The former task examines temporal-cultural feature recognition based on single- or multi-image inputs through multiple-choice visual question answering, while the latter focuses on transforming traditional attire into modern designs through cultural element inheritance and modern context adaptation. Our evaluation shows that closed VLMs perform comparably to non-experts on visual cutural understanding but fall short by 10\% to human experts, while open VLMs lags further behind non-experts. For the transcreation task, multi-faceted human evaluation indicates that the best-performing model achieves a success rate of only 42\%. Our benchmark provides an essential testbed, revealing significant challenges in this new direction of temporal cultural understanding and creative adaptation. | [
{
"version": "v1",
"created": "Mon, 02 Jun 2025 11:43:46 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 05:37:58 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Zhou",
"Li",
""
],
[
"Yu",
"Lutong",
""
],
[
"Xie",
"Dongchu",
""
],
[
"Cheng",
"Shaohuan",
""
],
[
"Li",
"Wenyan",
""
],
[
"Li",
"Haizhou",
""
]
] |
2506.01772 | Simon-Raphael Fischer | Simon-Raphael Fischer | Extension of algebroids Part I: The Construction | v2: fixing some typos and wrong labelling | null | null | null | math.DG | http://creativecommons.org/licenses/by/4.0/ | In this series of two papers we will generalise the concept of extending a Lie algebroid by a Lie algebra bundle, leading to a notion of extending a Lie algebroid by another Lie algebroid whose orbits lie in the orbits of the former algebroid. The resulting Lie algebroid's anchor will be the sum of the two initial anchors such that the constructions will be similar to matched pairs of Lie algebroids, but with the major difference that we will allow curvatures. In this part of this series we will focus on the canonical construction making use of strict covariant adjustments, a generalisation of Maurer-Cartan forms in the context of gauge theories equipped with a Lie groupoid action instead of a Lie group action. That is, a Cartan connection with certain conditions on the curvature. The second paper will introduce and explain the obstruction of the extension provided here. Examples will include locally split structures as in Poisson geometry.
As a side result we achieve strong hints towards a possible obstruction theory for certain Cartan connections on Lie algebroids, which will be related to the obstruction of (non-trivial) action algebroids; generalising the statement of the action algebroid structure induced by flat Cartan connections. | [
{
"version": "v1",
"created": "Mon, 02 Jun 2025 15:18:09 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 18:30:04 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Fischer",
"Simon-Raphael",
""
]
] |
2506.01857 | Boryeu Mao | Boryeu Mao | Protein folding classes -- High-dimensional geometry of amino acid composition space revisited | 48 pages, 6 figures, 4 tables | null | null | null | q-bio.BM q-bio.QM | http://creativecommons.org/licenses/by-nc-sa/4.0/ | In this study, the distributions of protein structure classes (or folding types) of experimentally determined structures from a legacy dataset and a comprehensive database SCOP are modeled precisely with geometric constructs such as convex polytopes in high-dimensional amino acid composition space. This is a follow-up of a previous non-statistical, geometry-motivated modeling of protein classes with ellipsoidal models, which are superseded presently in three important respects: (1) as a paradigm shift descriptive 'distribution model' of experimental data is de-coupled from, and serves as the basis for, possible future predictive 'domain model' generalizable to proteins in the same class for which 3D structures have yet to be determined experimentally, (2) the geometric and analytic characteristics of class distributions are obtained via exact computational geometry calculations, and (3) the full data from a comprehensive database are included in such calculations, eschewing training set selection and biases. In contrast to statistical and machine-learning approaches, the analytical, non-statistical geometry models of protein class distributions demonstrated in this study furnish complete and precise information on their size and relative disposition in the high-dimensional space (vis-\`a-vis any overlaps leading to ambiguity and limits in classification). Intended principally as accurate and summary description of the complex relationships between amino acid composition and protein classes, and suitably as a basis for predictive modeling where permissible, the results suggest that pen-ultimately they may be useful adjuncts for validating sequence-based protein structure predictions and contribute to theoretical and fundamental understanding of secondary structure formation and protein folding, demonstrating the role of high dimensional amino acid composition space in protein studies. | [
{
"version": "v1",
"created": "Mon, 02 Jun 2025 16:44:02 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 00:18:49 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Mao",
"Boryeu",
""
]
] |
2506.02156 | Xiaolin Li | Xiaolin Li, Ninghui Li, Boyang Wang, Wenhai Sun | Mitigating Data Poisoning Attacks to Local Differential Privacy | The 32nd ACM Conference on Computer and Communications Security (CCS 2025) | null | null | null | cs.CR | http://creativecommons.org/licenses/by/4.0/ | The distributed nature of local differential privacy (LDP) invites data poisoning attacks and poses unforeseen threats to the underlying LDP-supported applications. In this paper, we propose a comprehensive mitigation framework for popular frequency estimation, which contains a suite of novel defenses, including malicious user detection, attack pattern recognition, and damaged utility recovery. In addition to existing attacks, we explore new adaptive adversarial activities for our mitigation design. For detection, we present a new method to precisely identify bogus reports and thus LDP aggregation can be performed over the ``clean'' data. When the attack behavior becomes stealthy and direct filtering out malicious users is difficult, we further propose a detection that can effectively recognize hidden adversarial patterns, thus facilitating the decision-making of service providers. These detection methods require no additional data and attack information and incur minimal computational cost. Our experiment demonstrates their excellent performance and substantial improvement over previous work in various settings. In addition, we conduct an empirical analysis of LDP post-processing for corrupted data recovery and propose a new post-processing method, through which we reveal new insights into protocol recommendations in practice and key design principles for future research. | [
{
"version": "v1",
"created": "Mon, 02 Jun 2025 18:37:15 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 18:08:15 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Li",
"Xiaolin",
""
],
[
"Li",
"Ninghui",
""
],
[
"Wang",
"Boyang",
""
],
[
"Sun",
"Wenhai",
""
]
] |
2506.02237 | Joaquin Eduardo Reyes Gonzalez | Joaquin E. Reyes Gonzalez, Charles Zhang, Rainni K. Chen, John Y.T. Wei, Maureen J. Lagos | Atomic-scale mapping of interfacial phonon modes in epitaxial YBa2Cu3O7-{\delta} / (La,Sr)(Al,Ta)O3 thin films: The role of surface phonons | 17 pages, 4 figures | null | 10.1103/lxqy-m2y9 | null | cond-mat.supr-con cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | We investigate the behavior of phonons at the epitaxial interface between YBa2Cu3O7-{\delta} thin film and (La,Sr)(Al,Ta)O3 substrate using vibrational electron energy loss spectroscopy. Interfacial phonon modes with different degrees of scattering localization were identified. We find evidence that surface contributions from the surrounding environment can impose additional scattering modulation into local EELS measurements at the interface. A method to remove those contributions is then used to isolate the phonon information at the interface. This work unveils interfacial phonon modes in a high-Tc cuprate superconductor, that are not accessible with traditional phonon spectroscopy techniques, and provides a method for probing interfacial phonons in complex oxide heterostructures. | [
{
"version": "v1",
"created": "Mon, 02 Jun 2025 20:28:49 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Gonzalez",
"Joaquin E. Reyes",
""
],
[
"Zhang",
"Charles",
""
],
[
"Chen",
"Rainni K.",
""
],
[
"Wei",
"John Y. T.",
""
],
[
"Lagos",
"Maureen J.",
""
]
] |
2506.02390 | Qin Xie | Qin Xie, Qinghua Zhang, Shuyin Xia, Xinran Zhou, and Guoyin Wang | GAdaBoost: An Efficient and Robust AdaBoost Algorithm Based on Granular-Ball Structure | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Adaptive Boosting (AdaBoost) faces significant challenges posed by label noise, especially in multiclass classification tasks. Existing methods either lack mechanisms to handle label noise effectively or suffer from high computational costs due to redundant data usage. Inspired by granular computing, this paper proposes granular adaptive boosting (GAdaBoost), a novel two-stage framework comprising a data granulation stage and an adaptive boosting stage, to enhance efficiency and robustness under noisy conditions. To validate its feasibility, an extension of SAMME, termed GAdaBoost.SA, is proposed. Specifically, first, a granular-ball generation method is designed to compress data while preserving diversity and mitigating label noise. Second, the granular ball-based SAMME algorithm focuses on granular balls rather than individual samples, improving efficiency and reducing sensitivity to noise. Experimental results on some noisy datasets show that the proposed approach achieves superior robustness and efficiency compared with existing methods, demonstrating that this work effectively extends AdaBoost and SAMME. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 03:10:20 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 14:21:52 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Xie",
"Qin",
""
],
[
"Zhang",
"Qinghua",
""
],
[
"Xia",
"Shuyin",
""
],
[
"Zhou",
"Xinran",
""
],
[
"Wang",
"Guoyin",
""
]
] |
2506.03506 | Che-Yu Chen | Ismael Ayuso, Mariam Bouhmadi-L\'opez, Che-Yu Chen, Xiao Yan Chew, Konstantinos Dialektopoulos, Yen Chin Ong | Insights in $f(Q)$ cosmology: the relevance of the connection | 21 pages, 6 figures, references added | null | null | RIKEN-iTHEMS-Report-25 | gr-qc astro-ph.CO hep-ph hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We explore the role of the affine connection in $f(Q)$ gravity, a modified theory where gravity is governed by non-metricity within the symmetric teleparallel framework. Although the connection is constrained to be flat and torsionless, it is not uniquely determined by the metric, allowing for multiple physically distinct formulations. We analyze three such connections compatible with a homogeneous and isotropic universe to show that they yield markedly different cosmological dynamics, even under the same functional form of $f(Q)$. Using both analytical and numerical methods, including a Born-Infeld type model of $f(Q)$, we demonstrate that specific connections can resolve cosmological singularities like the Big Bang and Big Rip, replacing them with smooth de Sitter phases. Others retain singularities but with notable modifications in their behavior. These findings highlight the physical relevance of connection choice in $f(Q)$ gravity and its potential to address fundamental cosmological questions. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 02:48:55 GMT"
},
{
"version": "v2",
"created": "Fri, 13 Jun 2025 09:03:20 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Ayuso",
"Ismael",
""
],
[
"Bouhmadi-López",
"Mariam",
""
],
[
"Chen",
"Che-Yu",
""
],
[
"Chew",
"Xiao Yan",
""
],
[
"Dialektopoulos",
"Konstantinos",
""
],
[
"Ong",
"Yen Chin",
""
]
] |
2506.03625 | Yong Gao Chen | Yong-Gao Chen and Hui Zhu | The number of primes not in a numerical semigroup | 12 pages | null | null | null | math.NT | http://creativecommons.org/licenses/by-nc-nd/4.0/ | For two coprime positive integers $a$ and $b$,let $\pi^* (a, b)$ be the number of primes that cannot be represented as $au+bv$, where $u$ and $v$ are nonnegative integers. It is clear that $\pi^* (a, b)\le \pi (ab-a-b)$, where $\pi (x)$ denotes the number of primes not exceeding $x$. In this paper, we prove that $\pi^* (a, b)\ge 0.04\pi (ab-a-b)$ and pose following conjecture: $\pi^* (a, b)\ge \frac 12 \pi (ab-a-b)$. This conjecture is confirmed for $1\le a\le 10$. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 07:08:29 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 09:10:05 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Chen",
"Yong-Gao",
""
],
[
"Zhu",
"Hui",
""
]
] |
2506.03947 | Jemima Tabeart | Jemima M. Tabeart, Selime G\"urol, John W. Pearson and Anthony T. Weaver | Block Alpha-Circulant Preconditioners for All-at-Once Diffusion-Based Covariance Operators | 27 pages, 8 figures, 8 Tables | null | null | null | math.NA cs.NA | http://creativecommons.org/licenses/by/4.0/ | Covariance matrices are central to data assimilation and inverse methods derived from statistical estimation theory. Previous work has considered the application of an all-at-once diffusion-based representation of a covariance matrix operator in order to exploit inherent parallellism in the underlying problem. In this paper, we provide practical methods to apply block $\alpha$-circulant preconditioners to the all-at-once system for the case where the main diffusion operation matrix cannot be readily diagonalized using a discrete Fourier transform. Our new framework applies the block $\alpha$-circulant preconditioner approximately by solving an inner block diagonal problem via a choice of inner iterative approaches. Our first method applies Chebyshev semi-iteration to a symmetric positive definite matrix, shifted by a complex scaling of the identity. We extend theoretical results for Chebyshev semi-iteration in the symmetric positive definite setting, to obtain computable bounds on the asymptotic convergence factor for each of the complex sub-problems. The second approach transforms the complex sub-problem into a (generalized) saddle point system with real coefficients. Numerical experiments reveal that in the case of unlimited computational resources, both methods can match the iteration counts of the `best-case' block $\alpha$-circulant preconditioner. We also provide a practical adaptation to the nested Chebyshev approach, which improves performance in the case of a limited computational budget. Using an appropriate choice of $\alpha$ our new approaches are robust and efficient in terms of outer iterations and matrix--vector products. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:34:48 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 06:21:26 GMT"
},
{
"version": "v3",
"created": "Mon, 16 Jun 2025 18:23:53 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Tabeart",
"Jemima M.",
""
],
[
"Gürol",
"Selime",
""
],
[
"Pearson",
"John W.",
""
],
[
"Weaver",
"Anthony T.",
""
]
] |
2506.03999 | Viktor Nilsson | Viktor Nilsson, Pierre Nyquist | Large deviations for scaled families of Schr\"odinger bridges with reflection | null | null | null | null | math.PR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we show a large deviation principle for certain sequences of static Schr\"{o}dinger bridges, typically motivated by a scale-parameter decreasing towards zero, extending existing large deviation results to cover a wider range of reference processes. Our results provide a theoretical foundation for studying convergence of such Schr\"{o}dinger bridges to their limiting optimal transport plans. Within generative modeling, Schr\"{o}dinger bridges, or entropic optimal transport problems, constitute a prominent class of methods, in part because of their computational feasibility in high-dimensional settings. Recently, Bernton et al. established a large deviation principle, in the small-noise limit, for fixed-cost entropic optimal transport problems. In this paper, we address an open problem posed by Bernton et al. and extend their results to hold for Schr\"{o}dinger bridges associated with certain sequences of more general reference measures with enough regularity in a similar small-noise limit. These can be viewed as sequences of entropic optimal transport plans with non-fixed cost functions. Using a detailed analysis of the associated Skorokhod maps and transition densities, we show that the new large deviation results cover Schr\"{o}dinger bridges where the reference process is a reflected diffusion on bounded convex domains, corresponding to recently introduced model choices in the generative modeling literature. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 14:28:34 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 17:04:45 GMT"
},
{
"version": "v3",
"created": "Tue, 17 Jun 2025 16:39:28 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Nilsson",
"Viktor",
""
],
[
"Nyquist",
"Pierre",
""
]
] |
2506.04079 | Andre Martins | Pedro Henrique Martins, Jo\~ao Alves, Patrick Fernandes, Nuno M. Guerreiro, Ricardo Rei, Amin Farajian, Mateusz Klimaszewski, Duarte M. Alves, Jos\'e Pombal, Nicolas Boizard, Manuel Faysse, Pierre Colombo, Fran\c{c}ois Yvon, Barry Haddow, Jos\'e G. C. de Souza, Alexandra Birch, Andr\'e F. T. Martins | EuroLLM-9B: Technical Report | 56 pages | null | null | null | cs.CL cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | This report presents EuroLLM-9B, a large language model trained from scratch to support the needs of European citizens by covering all 24 official European Union languages and 11 additional languages. EuroLLM addresses the issue of European languages being underrepresented and underserved in existing open large language models. We provide a comprehensive overview of EuroLLM-9B's development, including tokenizer design, architectural specifications, data filtering, and training procedures. We describe the pre-training data collection and filtering pipeline, including the creation of EuroFilter, an AI-based multilingual filter, as well as the design of EuroBlocks-Synthetic, a novel synthetic dataset for post-training that enhances language coverage for European languages. Evaluation results demonstrate EuroLLM-9B's competitive performance on multilingual benchmarks and machine translation tasks, establishing it as the leading open European-made LLM of its size. To support open research and adoption, we release all major components of this work, including the base and instruction-tuned models, the EuroFilter classifier, and the synthetic post-training dataset. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 15:43:31 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 18:23:31 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Martins",
"Pedro Henrique",
""
],
[
"Alves",
"João",
""
],
[
"Fernandes",
"Patrick",
""
],
[
"Guerreiro",
"Nuno M.",
""
],
[
"Rei",
"Ricardo",
""
],
[
"Farajian",
"Amin",
""
],
[
"Klimaszewski",
"Mateusz",
""
],
[
"Alves",
"Duarte M.",
""
],
[
"Pombal",
"José",
""
],
[
"Boizard",
"Nicolas",
""
],
[
"Faysse",
"Manuel",
""
],
[
"Colombo",
"Pierre",
""
],
[
"Yvon",
"François",
""
],
[
"Haddow",
"Barry",
""
],
[
"de Souza",
"José G. C.",
""
],
[
"Birch",
"Alexandra",
""
],
[
"Martins",
"André F. T.",
""
]
] |
2506.04269 | Theodore Modis Dr. | Theodore Modis | Complexity in the Wake of Artificial Intelligence | null | COMPLEXITY 2025 | 10.1155/cplx/7656280 | null | physics.soc-ph physics.app-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This study aims to evaluate quantitatively, albeit in arbitrary units, the evolution of complexity of the human system since the domestication of fire. This is made possible by studying the timing of the 14 most important milestones, breaks in historical perspective, in the evolution of humans. AI is considered here as the latest such milestone with importance comparable to that of the Internet. The complexity is modeled to have evolved along a bell-shaped curve, reaching a maximum around our times, and soon entering a declining trajectory. According to this curve, the next evolutionary milestone of comparable importance is expected around 2050 and should add less complexity than AI but more than the milestone grouping together nuclear energy, DNA, and the transistor. The peak of the complexity curve coincides squarely with the life span of the baby boomers. The peak in the rate of growth of the world population precedes the complexity peak by 25 years, which is about the time it takes a young man or woman before they are able to add complexity to the human system in a significant way. It is in society's interest to flatten the complexity bell-shaped curve to whatever extent this is possible in order enjoy complexity longer. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 13:49:41 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 07:04:53 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Modis",
"Theodore",
""
]
] |
2506.04336 | Patricia Iglesias-Navarro | Patricia Iglesias-Navarro, Marc Huertas-Company, Pablo P\'erez-Gonz\'alez, Johan H. Knapen, ChangHoon Hahn, Anton M. Koekemoer, Steven L. Finkelstein, Natalia Villanueva, Andr\'es Asensio Ramos | Simulation-based inference of galaxy properties from JWST pixels | 13 pages, 7 figures. Submitted to A&A, comments welcome. Software available at https://github.com/patriglesias/SBIPIX.git | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | We present an efficient Bayesian SED-fitting framework tailored to multiwavelength pixel photometry from the JWST Advanced Deep Extragalactic Survey (JADES). Our method employs simulation-based inference to enable rapid posterior sampling across galaxy pixels, leveraging the unprecedented spatial resolution, wavelength coverage, and depth provided by the survey. It is trained on synthetic photometry generated from MILES stellar population models, incorporating both parametric and non-parametric SFHs, realistic noise, and JADES-like filter sensitivity thresholds. We validate this amortised inference approach on mock datasets, achieving robust and well-calibrated posterior distributions, with an $R^2$ score of 0.99 for stellar mass. Applying our pipeline to real observations, we derive spatially resolved maps of stellar population properties down to $\mathrm{S/N}_{\rm{pixel}}=5$ (averaged over F277W, F356W, F444W) for 1083 JADES galaxies and ~2 million pixels with spectroscopic redshifts. These maps enable the identification of dusty or starburst regions and offer insights into mass growth and the structural assembly. We assess the outshining phenomenon by comparing pixel-based and integrated stellar mass estimates, finding limited impact only in low-mass galaxies ($<10^8M_{\odot}$) but systematic differences of ~0.20 dex linked to SFH priors. With an average posterior sampling speed of $10^{-4}$ seconds per pixel and a total inference time of ~1 CPU-day for the full dataset, our model offers a scalable solution for extracting high-fidelity stellar population properties from HST+JWST datasets, opening the way for statistical studies at sub-galactic scales. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 18:00:19 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 11:56:40 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Iglesias-Navarro",
"Patricia",
""
],
[
"Huertas-Company",
"Marc",
""
],
[
"Pérez-González",
"Pablo",
""
],
[
"Knapen",
"Johan H.",
""
],
[
"Hahn",
"ChangHoon",
""
],
[
"Koekemoer",
"Anton M.",
""
],
[
"Finkelstein",
"Steven L.",
""
],
[
"Villanueva",
"Natalia",
""
],
[
"Ramos",
"Andrés Asensio",
""
]
] |
2506.06210 | Pavel Komarov | Pavel Komarov | Spectral Derivatives | Package at https://pypi.org/project/spectral-derivatives/ | null | null | null | eess.SP math.HO | http://creativecommons.org/licenses/by/4.0/ | One of the happiest accidents in all math is the ease of transforming a function to and taking derivatives in the Fourier frequency domain. But in order to exploit this extraordinary fact without serious artefacting, and in order to be able to use a computer, we need quite a bit of extra knowledge and care. This document sets out the math behind the spectral-derivatives Python package. I touch on fundamental signal processing and calculus concepts as necessary and build upwards. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 16:16:30 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 18:11:49 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Komarov",
"Pavel",
""
]
] |
2506.06224 | Cesare Vianello | Cesare Vianello, Matteo Ferraretto and Luca Salasnich | Finite-temperature entanglement and coherence in asymmetric bosonic Josephson junctions | 14 pages, 9 figures | Phys. Rev. A 111, 063310 (2025) | 10.1103/v6xk-3xsh | null | cond-mat.quant-gas quant-ph | http://creativecommons.org/licenses/by/4.0/ | We investigate the finite-temperature properties of a bosonic Josephson junction composed of N interacting atoms confined by a quasi-one-dimensional asymmetric double-well potential, modeled by the two-site Bose-Hubbard Hamiltonian. We compute numerically the spectral decomposition of the statistical ensemble of states, the thermodynamic and entanglement entropies, the population imbalance, the quantum Fisher information, and the coherence visibility. We analyze their dependence on the system parameters, showing in particular how finite temperature and on-site energy asymmetry affect the entanglement and coherence properties of the system. Moreover, starting from a quantum phase model which accurately describes the system over a wide range of interactions, we develop a reliable description of the strong tunneling regime, where thermal averages may be computed analytically using a modified Boltzmann weight involving an effective temperature. We discuss the possibility of applying this effective description to other models in suitable regimes. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 16:36:45 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Vianello",
"Cesare",
""
],
[
"Ferraretto",
"Matteo",
""
],
[
"Salasnich",
"Luca",
""
]
] |
2506.06290 | MingYu Lu | Mingyu Lu, Ethan Weinberger, Chanwoo Kim, Su-In Lee | CellCLIP -- Learning Perturbation Effects in Cell Painting via Text-Guided Contrastive Learning | null | null | null | null | cs.LG cs.AI cs.CV | http://creativecommons.org/licenses/by/4.0/ | High-content screening (HCS) assays based on high-throughput microscopy techniques such as Cell Painting have enabled the interrogation of cells' morphological responses to perturbations at an unprecedented scale. The collection of such data promises to facilitate a better understanding of the relationships between different perturbations and their effects on cellular state. Towards achieving this goal, recent advances in cross-modal contrastive learning could, in theory, be leveraged to learn a unified latent space that aligns perturbations with their corresponding morphological effects. However, the application of such methods to HCS data is not straightforward due to substantial differences in the semantics of Cell Painting images compared to natural images, and the difficulty of representing different classes of perturbations (e.g., small molecule vs CRISPR gene knockout) in a single latent space. In response to these challenges, here we introduce CellCLIP, a cross-modal contrastive learning framework for HCS data. CellCLIP leverages pre-trained image encoders coupled with a novel channel encoding scheme to better capture relationships between different microscopy channels in image embeddings, along with natural language encoders for representing perturbations. Our framework outperforms current open-source models, demonstrating the best performance in both cross-modal retrieval and biologically meaningful downstream tasks while also achieving significant reductions in computation time. | [
{
"version": "v1",
"created": "Fri, 16 May 2025 23:07:51 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 04:58:45 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Lu",
"Mingyu",
""
],
[
"Weinberger",
"Ethan",
""
],
[
"Kim",
"Chanwoo",
""
],
[
"Lee",
"Su-In",
""
]
] |
2506.06338 | Max Chicky Fang | Max Chicky Fang | Closed Form of a Generalized Sinkhorn Limit | 17 pages, 9 pages of Appendix | null | null | null | math.GM | http://creativecommons.org/licenses/by-nc-sa/4.0/ | The Kruithof iterative scaling process, which adjusts matrices to meet target row and column sums, is a longstanding problem that lacks a general closed form for its limit. While Nathanson derived the closed form for the Sinkhorn limit of $2\times 2$ matrices when target row and column sums are 1, and recent work by Rowland and Wu has advanced understanding of Sinkhorn limits for $3\times 3$, and general $n\times m$ matrices through polynomials, a "generalized Sinkhorn limit" (i.e. the original "Kruithof limit", with arbitrary target sums) remains elusive. Here, we derive the closed form for the generalized Sinkhorn limit of $2\times 2$ matrices, and discuss how this approach can be extended to larger matrices. More significantly, we prove that for any positive $n \times m$ matrix and positive target row and column sums, each entry in the generalized Sinkhorn limit is algebraic over the input data with degree at most $\binom{n+m-2}{n-1}$. | [
{
"version": "v1",
"created": "Sat, 31 May 2025 19:37:22 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 00:22:34 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Fang",
"Max Chicky",
""
]
] |
2506.06445 | Mayuko Mori | Mayuko Mori, Akihiko Fukui, Teruyuki Hirano, Norio Narita, John H. Livingston, Khalid Barkaoui, Karen A. Collins, Jerome P. de Leon, Kai Ikuta, Yugo Kawai, Richard P. Schwarz, Avi Shporer, Gregor Srdoc | Multi-band, Multi-epoch Photometry of the Spot-crossing System TOI-3884: Refined System Geometry and Spot Properties | 22 pages, 14 figures, accepted for publication in AJ. [ver. 2] This is a revised version of the manuscript incorporating corrections based on valuable feedback. In particular, the definition of stellar inclination has been updated to follow the conventional standard, resulting in revised values for related parameters. These corrections were made after acceptance but prior to journal publication | null | null | null | astro-ph.EP astro-ph.SR | http://creativecommons.org/licenses/by/4.0/ | Spot-crossing transits offer a unique opportunity to probe spot properties such as temperature, size, and surface distribution. TOI-3884 is a rare system in which spot-crossing features are persistently observed during every transit. This is due to its unusual configuration: a nearly poler orbit super-Neptune transits a pole-on mid-M dwarf, repeatedly crossing a polar spot. However, previous studies have reported discrepant values in key system parameters, such as stellar inclination and obliquity. To address this, we conducted multi-band, multi-epoch transit observations of TOI-3884b using the MuSCAT instrument series, along with photometric monitoring with the LCO 1m telescopes/Sinistro. We detected time-dependent variations in the spot-crossing signals, indicating that the spot is not exactly on the pole. From the monitoring data, we measured a stellar rotation period of $11.043~_{-0.053}^{+0.054}$ days with a modulation amplitude of $\sim$5% in the r-band, consistent with the time variability in the spot-crossing features. Our analysis reconciles previous discrepancies and improves the constraints on the parameters of the system geometry ($i_\star = 139.9~^{+1.2}_{-2.0}$ deg and $\lambda =41.0~^{+3.7}_{-9.0}$ deg) and those of the spot properties (spot radius of $0.425~_{-0.011}^{+0.018}~R_\star$ and spot-photosphere temperature difference of $200~_{-9}^{+11}$ K). These results provide a critical context for interpreting upcoming transmission spectroscopy of TOI-3884b, as well as yielding new insights into the magnetic activity and spin-orbit geometry of M dwarfs. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 18:07:55 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 07:48:42 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Mori",
"Mayuko",
""
],
[
"Fukui",
"Akihiko",
""
],
[
"Hirano",
"Teruyuki",
""
],
[
"Narita",
"Norio",
""
],
[
"Livingston",
"John H.",
""
],
[
"Barkaoui",
"Khalid",
""
],
[
"Collins",
"Karen A.",
""
],
[
"de Leon",
"Jerome P.",
""
],
[
"Ikuta",
"Kai",
""
],
[
"Kawai",
"Yugo",
""
],
[
"Schwarz",
"Richard P.",
""
],
[
"Shporer",
"Avi",
""
],
[
"Srdoc",
"Gregor",
""
]
] |
2506.06499 | Alex Havrilla | Alex Havrilla, Edward Hughes, Mikayel Samvelyan, Jacob Abernethy | SPARQ: Synthetic Problem Generation for Reasoning via Quality-Diversity Algorithms | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Large language model (LLM) driven synthetic data generation has emerged as a powerful method for improving model reasoning capabilities. However, most methods either distill large state-of-the-art models into small students or use natural ground-truth problem statements to guarantee problem statement quality. This limits the scalability of these approaches to more complex and diverse problem domains. To address this, we present SPARQ: Synthetic Problem Generation for Reasoning via Quality-Diversity Algorithms, a novel approach for generating high-quality and diverse synthetic math problem and solution pairs using only a single model by measuring a problem's solve-rate: a proxy for problem difficulty. Starting from a seed dataset of 7.5K samples, we generate over 20 million new problem-solution pairs. We show that filtering the generated data by difficulty and then fine-tuning the same model on the resulting data improves relative model performance by up to 24\%. Additionally, we conduct ablations studying the impact of synthetic data quantity, quality and diversity on model generalization. We find that higher quality, as measured by problem difficulty, facilitates better in-distribution performance. Further, while generating diverse synthetic data does not as strongly benefit in-distribution performance, filtering for more diverse data facilitates more robust OOD generalization. We also confirm the existence of model and data scaling laws for synthetically generated problems, which positively benefit downstream model generalization. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 19:49:42 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 12:55:30 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Havrilla",
"Alex",
""
],
[
"Hughes",
"Edward",
""
],
[
"Samvelyan",
"Mikayel",
""
],
[
"Abernethy",
"Jacob",
""
]
] |
2506.06640 | Haijun Li | Haijun Li | The combinatorics of identities involving overpartitions with distinct parts | 14 pages | null | null | null | math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recently, Andrews and EI Bachraoui discovered several companions for some famous $q$-series formulas, and derived some new identities involving partitions and overpartitions with distinct parts. In this paper, we shall refine their results by the number of parts of partitions and furthermore, we will also provide the combinatorial proofs for those partition identities. | [
{
"version": "v1",
"created": "Sat, 07 Jun 2025 03:28:04 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 02:38:56 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Li",
"Haijun",
""
]
] |
2506.06654 | Bingyan Han | Erhan Bayraktar and Bingyan Han | Goal-based portfolio selection with mental accounting | 68 pages, 8 figures | null | null | null | q-fin.PM math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a continuous-time portfolio selection framework that reflects goal-based investment principles and mental accounting behavior. In this framework, an investor with multiple investment goals constructs separate portfolios, each corresponding to a specific goal, with penalties imposed on fund transfers between these goals, referred to as mental costs. By applying the stochastic Perron's method, we demonstrate that the value function is the unique constrained viscosity solution of a Hamilton-Jacobi-Bellman equation system. Numerical analysis reveals several key features: the free boundaries exhibit complex shapes with bulges and notches; the optimal strategy for one portfolio depends on the wealth level of another; investors must diversify both among stocks and across portfolios; and they may postpone reallocating surplus from an important goal to a less important one until the former's deadline approaches. | [
{
"version": "v1",
"created": "Sat, 07 Jun 2025 04:16:44 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 06:22:13 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Bayraktar",
"Erhan",
""
],
[
"Han",
"Bingyan",
""
]
] |
2506.07004 | Ye-Ji Mun | Zhe Huang, Ye-Ji Mun, Fatemeh Cheraghi Pouria, Katherine Driggs-Campbell | Hierarchical Intention Tracking with Switching Trees for Real-Time Adaptation to Dynamic Human Intentions during Collaboration | 15 pages, 10 figures | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | During collaborative tasks, human behavior is guided by multiple levels of intentions that evolve over time, such as task sequence preferences and interaction strategies. To adapt to these changing preferences and promptly correct any inaccurate estimations, collaborative robots must accurately track these dynamic human intentions in real time. We propose a Hierarchical Intention Tracking (HIT) algorithm for collaborative robots to track dynamic and hierarchical human intentions effectively in real time. HIT represents human intentions as intention trees with arbitrary depth, and probabilistically tracks human intentions by Bayesian filtering, upward measurement propagation, and downward posterior propagation across all levels. We develop a HIT-based robotic system that dynamically switches between Interaction-Task and Verification-Task trees for a collaborative assembly task, allowing the robot to effectively coordinate human intentions at three levels: task-level (subtask goal locations), interaction-level (mode of engagement with the robot), and verification-level (confirming or correcting intention recognition). Our user study shows that our HIT-based collaborative robot system surpasses existing collaborative robot solutions by achieving a balance between efficiency, physical workload, and user comfort while ensuring safety and task completion. Post-experiment surveys further reveal that the HIT-based system enhances the user trust and minimizes interruptions to user's task flow through its effective understanding of human intentions across multiple levels. | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 05:30:36 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 05:52:09 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Huang",
"Zhe",
""
],
[
"Mun",
"Ye-Ji",
""
],
[
"Pouria",
"Fatemeh Cheraghi",
""
],
[
"Driggs-Campbell",
"Katherine",
""
]
] |
2506.07686 | Anatolii Kononov Dr. | Anatolii Kononov | An Efficient Method for Evaluating the Feasibility of Spaceborne SAR for Ocean Ship Detection | 5 pages, 4 figures, 2 tables | null | null | null | eess.SP | http://creativecommons.org/licenses/by/4.0/ | This letter presents an effective method for assessing the feasibility of detecting ocean ships using spaceborne synthetic aperture radar (SAR). The technique employs the minimum detectable radar cross-section criterion under specified false alarm and detection probabilities. The benefits of the proposed method are illustrated by evaluating the feasibility of detecting small ships with SAR from a satellite in very low Earth orbit. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 12:16:16 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 10:48:38 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Kononov",
"Anatolii",
""
]
] |
2506.07801 | Iustin Sirbu | Iustin Sirbu (1), Robert-Adrian Popovici (1), Cornelia Caragea (2), Stefan Trausan-Matu (1), Traian Rebedea (1 and 3) ((1) National University of Science and Technology POLITEHNICA Bucharest, (2) University of Illinois Chicago, (3) NVIDIA) | MultiMatch: Multihead Consistency Regularization Matching for Semi-Supervised Text Classification | null | null | null | null | cs.CL cs.AI cs.LG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | We introduce MultiMatch, a novel semi-supervised learning (SSL) algorithm combining the paradigms of co-training and consistency regularization with pseudo-labeling. At its core, MultiMatch features a three-fold pseudo-label weighting module designed for three key purposes: selecting and filtering pseudo-labels based on head agreement and model confidence, and weighting them according to the perceived classification difficulty. This novel module enhances and unifies three existing techniques -- heads agreement from Multihead Co-training, self-adaptive thresholds from FreeMatch, and Average Pseudo-Margins from MarginMatch -- resulting in a holistic approach that improves robustness and performance in SSL settings. Experimental results on benchmark datasets highlight the superior performance of MultiMatch, achieving state-of-the-art results on 9 out of 10 setups from 5 natural language processing datasets and ranking first according to the Friedman test among 19 methods. Furthermore, MultiMatch demonstrates exceptional robustness in highly imbalanced settings, outperforming the second-best approach by 3.26% -- and data imbalance is a key factor for many text classification tasks. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 14:27:47 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 20:28:26 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Sirbu",
"Iustin",
"",
"National University of Science and Technology POLITEHNICA Bucharest"
],
[
"Popovici",
"Robert-Adrian",
"",
"National University of Science and Technology POLITEHNICA Bucharest"
],
[
"Caragea",
"Cornelia",
"",
"University of Illinois Chicago"
],
[
"Trausan-Matu",
"Stefan",
"",
"National University of Science and Technology POLITEHNICA Bucharest"
],
[
"Rebedea",
"Traian",
"",
"1 and 3"
]
] |
2506.07838 | Fr\'ed\'eric Suter | Fr\'ed\'eric Suter, Tain\~a Coleman, \.Ilkay Altinta\c{s}, Rosa M. Badia, Bartosz Balis, Kyle Chard, Iacopo Colonnelli, Ewa Deelman, Paolo Di Tommaso, Thomas Fahringer, Carole Goble, Shantenu Jha, Daniel S. Katz, Johannes K\"oster, Ulf Leser, Kshitij Mehta, Hilary Oliver, J.-Luc Peterson, Giovanni Pizzi, Lo\"ic Pottier, Ra\"ul Sirvent, Eric Suchyta, Douglas Thain, Sean R. Wilkinson, Justin M. Wozniak, Rafael Ferreira da Silva | A Terminology for Scientific Workflow Systems | null | null | null | null | cs.DC | http://creativecommons.org/licenses/by/4.0/ | The term scientific workflow has evolved over the last two decades to encompass a broad range of compositions of interdependent compute tasks and data movements. It has also become an umbrella term for processing in modern scientific applications. Today, many scientific applications can be considered as workflows made of multiple dependent steps, and hundreds of workflow management systems (WMSs) have been developed to manage and run these workflows. However, no turnkey solution has emerged to address the diversity of scientific processes and the infrastructure on which they are implemented. Instead, new research problems requiring the execution of scientific workflows with some novel feature often lead to the development of an entirely new WMS. A direct consequence is that many existing WMSs share some salient features, offer similar functionalities, and can manage the same categories of workflows but also have some distinct capabilities. This situation makes researchers who develop workflows face the complex question of selecting a WMS. This selection can be driven by technical considerations, to find the system that is the most appropriate for their application and for the resources available to them, or other factors such as reputation, adoption, strong community support, or long-term sustainability. To address this problem, a group of WMS developers and practitioners joined their efforts to produce a community-based terminology of WMSs. This paper summarizes their findings and introduces this new terminology to characterize WMSs. This terminology is composed of fives axes: workflow characteristics, composition, orchestration, data management, and metadata capture. Each axis comprises several concepts that capture the prominent features of WMSs. Based on this terminology, this paper also presents a classification of 23 existing WMSs according to the proposed axes and terms. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 15:04:48 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 13:42:31 GMT"
},
{
"version": "v3",
"created": "Mon, 16 Jun 2025 16:40:45 GMT"
},
{
"version": "v4",
"created": "Mon, 16 Jun 2025 18:10:41 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Suter",
"Frédéric",
""
],
[
"Coleman",
"Tainã",
""
],
[
"Altintaş",
"İlkay",
""
],
[
"Badia",
"Rosa M.",
""
],
[
"Balis",
"Bartosz",
""
],
[
"Chard",
"Kyle",
""
],
[
"Colonnelli",
"Iacopo",
""
],
[
"Deelman",
"Ewa",
""
],
[
"Di Tommaso",
"Paolo",
""
],
[
"Fahringer",
"Thomas",
""
],
[
"Goble",
"Carole",
""
],
[
"Jha",
"Shantenu",
""
],
[
"Katz",
"Daniel S.",
""
],
[
"Köster",
"Johannes",
""
],
[
"Leser",
"Ulf",
""
],
[
"Mehta",
"Kshitij",
""
],
[
"Oliver",
"Hilary",
""
],
[
"Peterson",
"J. -Luc",
""
],
[
"Pizzi",
"Giovanni",
""
],
[
"Pottier",
"Loïc",
""
],
[
"Sirvent",
"Raül",
""
],
[
"Suchyta",
"Eric",
""
],
[
"Thain",
"Douglas",
""
],
[
"Wilkinson",
"Sean R.",
""
],
[
"Wozniak",
"Justin M.",
""
],
[
"da Silva",
"Rafael Ferreira",
""
]
] |
2506.07917 | Allen Tu | Allen Tu, Haiyang Ying, Alex Hanson, Yonghan Lee, Tom Goldstein, Matthias Zwicker | Speedy Deformable 3D Gaussian Splatting: Fast Rendering and Compression of Dynamic Scenes | Project Page: https://speede3dgs.github.io/ | Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025, pp. 21537-21546 | null | null | cs.GR cs.CV | http://creativecommons.org/licenses/by/4.0/ | Recent extensions of 3D Gaussian Splatting (3DGS) to dynamic scenes achieve high-quality novel view synthesis by using neural networks to predict the time-varying deformation of each Gaussian. However, performing per-Gaussian neural inference at every frame poses a significant bottleneck, limiting rendering speed and increasing memory and compute requirements. In this paper, we present Speedy Deformable 3D Gaussian Splatting (SpeeDe3DGS), a general pipeline for accelerating the rendering speed of dynamic 3DGS and 4DGS representations by reducing neural inference through two complementary techniques. First, we propose a temporal sensitivity pruning score that identifies and removes Gaussians with low contribution to the dynamic scene reconstruction. We also introduce an annealing smooth pruning mechanism that improves pruning robustness in real-world scenes with imprecise camera poses. Second, we propose GroupFlow, a motion analysis technique that clusters Gaussians by trajectory similarity and predicts a single rigid transformation per group instead of separate deformations for each Gaussian. Together, our techniques accelerate rendering by $10.37\times$, reduce model size by $7.71\times$, and shorten training time by $2.71\times$ on the NeRF-DS dataset. SpeeDe3DGS also improves rendering speed by $4.20\times$ and $58.23\times$ on the D-NeRF and HyperNeRF vrig datasets. Our methods are modular and can be integrated into any deformable 3DGS or 4DGS framework. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 16:30:48 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Tu",
"Allen",
""
],
[
"Ying",
"Haiyang",
""
],
[
"Hanson",
"Alex",
""
],
[
"Lee",
"Yonghan",
""
],
[
"Goldstein",
"Tom",
""
],
[
"Zwicker",
"Matthias",
""
]
] |
2506.07934 | Juan Salvador G\'omez Casanueva | Juan S. G\'omez | Codimension-Two Spacelike Submanifolds with Umbilical Lightlike Normal Sections and Their Relationship to Lightlike Hypersurfaces | 35 pages | null | null | null | math.DG | http://creativecommons.org/licenses/by/4.0/ | We study codimension-two spacelike submanifolds in Lorentzian spacetimes that admit umbilical lightlike normal directions. We show that such submanifolds are subject to strong geometric and topological constraints, establishing explicit relationships between extrinsic geometry, mean curvature, and shear-isotropy. In the compact case, we obtain sharp restrictions on their topology. We precisely characterize when the umbilical lightlike normal vector field can be rescaled to be parallel, in terms of the curvature tensor of the ambient spacetime, and prove that this property is conformally invariant. Our main result is a factorization theorem: any such submanifold is contained in a lightlike hypersurface, which is totally umbilical whenever the lightlike normal direction is umbilical. We also provide explicit conformal relations between the induced metrics on the family of spacelike leaves generated by the lightlike normal flow, with consequences for isometry, parallelism of lightlike normal directions, volume evolution, and variational properties. These results yield a detailed geometric framework relevant to the mathematical study of horizons and lightlike structures in general relativity. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 16:54:34 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 17:10:03 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Gómez",
"Juan S.",
""
]
] |
2506.08001 | Weiyang Liu | Zeju Qiu, Simon Buchholz, Tim Z. Xiao, Maximilian Dax, Bernhard Sch\"olkopf, Weiyang Liu | Reparameterized LLM Training via Orthogonal Equivalence Transformation | Technical report v3 (38 pages, 26 figures, project page: https://spherelab.ai/poet/, v3: added singular spectrum and energy analyses in Section 4) | null | null | null | cs.LG cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While large language models (LLMs) are driving the rapid advancement of artificial intelligence, effectively and reliably training these large models remains one of the field's most significant challenges. To address this challenge, we propose POET, a novel reParameterized training algorithm that uses Orthogonal Equivalence Transformation to optimize neurons. Specifically, POET reparameterizes each neuron with two learnable orthogonal matrices and a fixed random weight matrix. Because of its provable preservation of spectral properties of weight matrices, POET can stably optimize the objective function with improved generalization. We further develop efficient approximations that make POET flexible and scalable for training large-scale neural networks. Extensive experiments validate the effectiveness and scalability of POET in training LLMs. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 17:59:34 GMT"
},
{
"version": "v2",
"created": "Sat, 14 Jun 2025 16:13:08 GMT"
},
{
"version": "v3",
"created": "Tue, 17 Jun 2025 16:44:36 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Qiu",
"Zeju",
""
],
[
"Buchholz",
"Simon",
""
],
[
"Xiao",
"Tim Z.",
""
],
[
"Dax",
"Maximilian",
""
],
[
"Schölkopf",
"Bernhard",
""
],
[
"Liu",
"Weiyang",
""
]
] |
2506.08026 | Xibai Wang | Xibai Wang | TIP-Search: Time-Predictable Inference Scheduling for Market Prediction under Uncertain Load | null | null | null | null | cs.AI cs.LG cs.SY eess.SY q-fin.CP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper proposes TIP-Search, a time-predictable inference scheduling framework for real-time market prediction under uncertain workloads. Motivated by the strict latency demands in high-frequency financial systems, TIP-Search dynamically selects a deep learning model from a heterogeneous pool, aiming to maximize predictive accuracy while satisfying per-task deadline constraints. Our approach profiles latency and generalization performance offline, then performs online task-aware selection without relying on explicit input domain labels. We evaluate TIP-Search on three real-world limit order book datasets (FI-2010, Binance BTC/USDT, LOBSTER AAPL) and demonstrate that it outperforms static baselines with up to 8.5% improvement in accuracy and 100% deadline satisfaction. Our results highlight the effectiveness of TIP-Search in robust low-latency financial inference under uncertainty. | [
{
"version": "v1",
"created": "Fri, 30 May 2025 14:52:01 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 19:58:59 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Wang",
"Xibai",
""
]
] |
2506.08273 | Bart{\l}omiej Dyda | Bart{\l}omiej Dyda | Fractional and non-fractional Hardy inequality on a lattice $\Z^d$ | 15 pages | null | null | null | math.FA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present simple proofs of a discrete fractional and non-fractional Hardy inequality, Our constants are explicit, but not optimal. In the class of power weights, we get a complete picture of when the non-fractional Hardy inequality holds, for any dimension $d$ of the lattice $\Z^d$ and exponent $0<p<\infty$. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 22:28:26 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 19:38:07 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Dyda",
"Bartłomiej",
""
]
] |
2506.08673 | Kushagra Chatterjee | Diptarka Chakraborty, Kushagra Chatterjee, Debarati Das, Tien Long Nguyen, Romina Nobahari | Towards Fair Representation: Clustering and Consensus | The paper has been accepted at the Conference on Learning Theory (COLT) 2025. We have fixed some typos in the theorem statements | null | null | null | cs.LG cs.DS | http://creativecommons.org/licenses/by/4.0/ | Consensus clustering, a fundamental task in machine learning and data analysis, aims to aggregate multiple input clusterings of a dataset, potentially based on different non-sensitive attributes, into a single clustering that best represents the collective structure of the data. In this work, we study this fundamental problem through the lens of fair clustering, as introduced by Chierichetti et al. [NeurIPS'17], which incorporates the disparate impact doctrine to ensure proportional representation of each protected group in the dataset within every cluster. Our objective is to find a consensus clustering that is not only representative but also fair with respect to specific protected attributes. To the best of our knowledge, we are the first to address this problem and provide a constant-factor approximation.
As part of our investigation, we examine how to minimally modify an existing clustering to enforce fairness -- an essential postprocessing step in many clustering applications that require fair representation. We develop an optimal algorithm for datasets with equal group representation and near-linear time constant factor approximation algorithms for more general scenarios with different proportions of two group sizes. We complement our approximation result by showing that the problem is NP-hard for two unequal-sized groups. Given the fundamental nature of this problem, we believe our results on Closest Fair Clustering could have broader implications for other clustering problems, particularly those for which no prior approximation guarantees exist for their fair variants. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 10:33:21 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 21:53:50 GMT"
},
{
"version": "v3",
"created": "Tue, 17 Jun 2025 13:14:47 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Chakraborty",
"Diptarka",
""
],
[
"Chatterjee",
"Kushagra",
""
],
[
"Das",
"Debarati",
""
],
[
"Nguyen",
"Tien Long",
""
],
[
"Nobahari",
"Romina",
""
]
] |
2506.08697 | Jacopo Somaglia | Dario Daniele Monticelli, Fabio Punzo, Jacopo Somaglia | Nonexistence results for the semilinear wave equation on graphs | null | null | null | null | math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate the semilinear wave equation with potential on weighted graphs. We establish sufficient conditions for the nonexistence of global-in-time solutions. Both nonnegative and sign-changing solutions are considered. In particular, the proof for sign-changing solutions relies on a novel technique for this type of result. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 11:24:51 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 20:43:59 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Monticelli",
"Dario Daniele",
""
],
[
"Punzo",
"Fabio",
""
],
[
"Somaglia",
"Jacopo",
""
]
] |
2506.08915 | Ananthu Aniraj | Ananthu Aniraj, Cassio F. Dantas, Dino Ienco, Diego Marcos | Inherently Faithful Attention Maps for Vision Transformers | null | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | We introduce an attention-based method that uses learned binary attention masks to ensure that only attended image regions influence the prediction. Context can strongly affect object perception, sometimes leading to biased representations, particularly when objects appear in out-of-distribution backgrounds. At the same time, many image-level object-centric tasks require identifying relevant regions, often requiring context. To address this conundrum, we propose a two-stage framework: stage 1 processes the full image to discover object parts and identify task-relevant regions, while stage 2 leverages input attention masking to restrict its receptive field to these regions, enabling a focused analysis while filtering out potentially spurious information. Both stages are trained jointly, allowing stage 2 to refine stage 1. Extensive experiments across diverse benchmarks demonstrate that our approach significantly improves robustness against spurious correlations and out-of-distribution backgrounds. Code: https://github.com/ananthu-aniraj/ifam | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 15:41:22 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 08:52:37 GMT"
},
{
"version": "v3",
"created": "Tue, 17 Jun 2025 13:45:06 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Aniraj",
"Ananthu",
""
],
[
"Dantas",
"Cassio F.",
""
],
[
"Ienco",
"Dino",
""
],
[
"Marcos",
"Diego",
""
]
] |
2506.09081 | Zheqi He | Zheqi He, Yesheng Liu, Jing-shu Zheng, Xuejing Li, Jin-Ge Yao, Bowen Qin, Richeng Xuan, Xi Yang | FlagEvalMM: A Flexible Framework for Comprehensive Multimodal Model Evaluation | null | null | null | null | cs.CV cs.AI cs.CL | http://creativecommons.org/licenses/by/4.0/ | We present FlagEvalMM, an open-source evaluation framework designed to comprehensively assess multimodal models across a diverse range of vision-language understanding and generation tasks, such as visual question answering, text-to-image/video generation, and image-text retrieval. We decouple model inference from evaluation through an independent evaluation service, thus enabling flexible resource allocation and seamless integration of new tasks and models. Moreover, FlagEvalMM utilizes advanced inference acceleration tools (e.g., vLLM, SGLang) and asynchronous data loading to significantly enhance evaluation efficiency. Extensive experiments show that FlagEvalMM offers accurate and efficient insights into model strengths and limitations, making it a valuable tool for advancing multimodal research. The framework is publicly accessible athttps://github.com/flageval-baai/FlagEvalMM. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 04:19:02 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 07:54:37 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"He",
"Zheqi",
""
],
[
"Liu",
"Yesheng",
""
],
[
"Zheng",
"Jing-shu",
""
],
[
"Li",
"Xuejing",
""
],
[
"Yao",
"Jin-Ge",
""
],
[
"Qin",
"Bowen",
""
],
[
"Xuan",
"Richeng",
""
],
[
"Yang",
"Xi",
""
]
] |
2506.09172 | Harshvardhan Sikka | Pranav Guruprasad, Yangyue Wang, Sudipta Chowdhury, Jaewoo Song, Harshvardhan Sikka | An Open-Source Software Toolkit & Benchmark Suite for the Evaluation and Adaptation of Multimodal Action Models | ICML CodeML Workshop, 13 Pages, 6 Figures, 2 Tables | null | null | null | cs.LG cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent innovations in multimodal action models represent a promising direction for developing general-purpose agentic systems, combining visual understanding, language comprehension, and action generation. We introduce MultiNet - a novel, fully open-source benchmark and surrounding software ecosystem designed to rigorously evaluate and adapt models across vision, language, and action domains. We establish standardized evaluation protocols for assessing vision-language models (VLMs) and vision-language-action models (VLAs), and provide open source software to download relevant data, models, and evaluations. Additionally, we provide a composite dataset with over 1.3 trillion tokens of image captioning, visual question answering, commonsense reasoning, robotic control, digital game-play, simulated locomotion/manipulation, and many more tasks. The MultiNet benchmark, framework, toolkit, and evaluation harness have been used in downstream research on the limitations of VLA generalization. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 18:38:19 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 03:30:48 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Guruprasad",
"Pranav",
""
],
[
"Wang",
"Yangyue",
""
],
[
"Chowdhury",
"Sudipta",
""
],
[
"Song",
"Jaewoo",
""
],
[
"Sikka",
"Harshvardhan",
""
]
] |
2506.09250 | Alex Lawsen | A. Lawsen | Comment on The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity | Comment on: arXiv:2506.06941 Latest version removes Claude as a co-author, in line with arXiv policies, it also corrects mistakes in sections 4 and 6 of the original submission, as well as several typographical errors | null | null | null | cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Shojaee et al. (2025) report that Large Reasoning Models (LRMs) exhibit "accuracy collapse" on planning puzzles beyond certain complexity thresholds. We demonstrate that their findings primarily reflect experimental design limitations rather than fundamental reasoning failures. Our analysis reveals three critical issues: (1) Tower of Hanoi experiments risk exceeding model output token limits, with models explicitly acknowledging these constraints in their outputs; (2) The authors' automated evaluation framework fails to distinguish between reasoning failures and practical constraints, leading to misclassification of model capabilities; (3) Most concerningly, their River Crossing benchmarks include mathematically impossible instances for N > 5 due to insufficient boat capacity, yet models are scored as failures for not solving these unsolvable problems. When we control for these experimental artifacts, by requesting generating functions instead of exhaustive move lists, preliminary experiments across multiple models indicate high accuracy on Tower of Hanoi instances previously reported as complete failures. These findings highlight the importance of careful experimental design when evaluating AI reasoning capabilities. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 21:16:53 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 18:04:21 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Lawsen",
"A.",
""
]
] |
2506.09265 | Thuc Mai | Thuc T. Mai, Amber McCreary, K.F. Garrity, Rebecca L. Dally, Sambridhi Shah, Bryan C. Chakoumakos, Md Nasim Afroj Taj, Jeffrey W. Lynn, Michael A. McGuire, Benjamin S. Conner, Mona Zebarjadi, Janice L. Musfeldt, Angela R. Hight Walker, Rahul Rao, Michael A. Susner | Spin-lattice entanglement in $\mathbf{CoPS}_3$ | 12 pages, 5 figures | null | null | null | cond-mat.str-el cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Complex chalcogenides in the $M$PS$_3$ family of materials ($M$ = Mn, Fe, Co, and Ni) display remarkably different phase progressions depending upon the metal center orbital filling, character of the P-P linkage, and size of the van der Waals gap. There is also a stacking pattern and spin state difference between the lighter and heavier transition metal-containing systems that places CoPS$_3$ at the nexus of these activities. Despite these unique properties, this compound is under-explored. Here, we bring together Raman scattering spectroscopy and infrared absorption spectroscopy with X-ray techniques to identify a structural component to the 119 K magnetic ordering transition as well as a remarkable lower temperature set of magnon-phonon pairs that engage in avoided crossings along with a magnetic scattering continuum that correlates with phonon lifetime effects. These findings point to strong spin-phonon entanglement as well as opportunities to control these effects under external stimuli. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 21:59:21 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 19:00:45 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Mai",
"Thuc T.",
""
],
[
"McCreary",
"Amber",
""
],
[
"Garrity",
"K. F.",
""
],
[
"Dally",
"Rebecca L.",
""
],
[
"Shah",
"Sambridhi",
""
],
[
"Chakoumakos",
"Bryan C.",
""
],
[
"Taj",
"Md Nasim Afroj",
""
],
[
"Lynn",
"Jeffrey W.",
""
],
[
"McGuire",
"Michael A.",
""
],
[
"Conner",
"Benjamin S.",
""
],
[
"Zebarjadi",
"Mona",
""
],
[
"Musfeldt",
"Janice L.",
""
],
[
"Walker",
"Angela R. Hight",
""
],
[
"Rao",
"Rahul",
""
],
[
"Susner",
"Michael A.",
""
]
] |
2506.09310 | Guo-He Li | Guo-He Li, Yeqi Fang, Yuchi Wu, and Jun Tao | Wormhole Solutions and pre-inflationary in $F(R, T)$ Gravity with Axion Fields | 15 pages, 9 figures | null | null | null | gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this study, we investigate wormhole solutions and inflationary initial conditions in the coupled axion-inflaton/scalar field system within $F(R,T)$ gravity. Notably, the Euclidean action is reduced by approximately $10^5$ compared with the GR case. In Euclidean AdS spacetime, we construct Euclidean (semi-)wormhole geometries that naturally set inflationary initial conditions. To enhance the probability of sustained inflation and address the short inflation duration issue in the no-boundary proposal, we constrain $\lambda$ to $\lambda < -\kappa$ by analyzing the properties of the Euclidean action. This constraint causes the probability weight $ P \propto e^{-S_E} $ to favor high-potential regions, achieving sufficient inflation while maintaining a nearly Gaussian perturbation spectrum. The results suggest that matter-geometry coupling in $F(R,T)$ gravity provides a novel mechanism for reconciling quantum cosmology with observational requirements. | [
{
"version": "v1",
"created": "Wed, 11 Jun 2025 00:32:34 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 14:12:21 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Li",
"Guo-He",
""
],
[
"Fang",
"Yeqi",
""
],
[
"Wu",
"Yuchi",
""
],
[
"Tao",
"Jun",
""
]
] |
2506.09316 | Yeonju Ro | Yeonju Ro, Zhenyu Zhang, Souvik Kundu, Zhangyang Wang, Aditya Akella | On-the-Fly Adaptive Distillation of Transformer to Dual-State Linear Attention | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Large language models (LLMs) excel at capturing global token dependencies via self-attention but face prohibitive compute and memory costs on lengthy inputs. While sub-quadratic methods (e.g., linear attention) can reduce these costs, they often degrade accuracy due to overemphasizing recent tokens. In this work, we first propose dual-state linear attention (DSLA), a novel design that maintains two specialized hidden states-one for preserving historical context and one for tracking recency-thereby mitigating the short-range bias typical of linear-attention architectures. To further balance efficiency and accuracy under dynamic workload conditions, we introduce DSLA-Serve, an online adaptive distillation framework that progressively replaces Transformer layers with DSLA layers at inference time, guided by a sensitivity-based layer ordering. DSLA-Serve uses a chained fine-tuning strategy to ensure that each newly converted DSLA layer remains consistent with previously replaced layers, preserving the overall quality. Extensive evaluations on commonsense reasoning, long-context QA, and text summarization demonstrate that DSLA-Serve yields 2.3x faster inference than Llama2-7B and 3.0x faster than the hybrid Zamba-7B, while retaining comparable performance across downstream tasks. Our ablation studies show that DSLA's dual states capture both global and local dependencies, addressing the historical-token underrepresentation seen in prior linear attentions. Codes are available at https://github.com/utnslab/DSLA-Serve. | [
{
"version": "v1",
"created": "Wed, 11 Jun 2025 01:25:06 GMT"
},
{
"version": "v2",
"created": "Thu, 12 Jun 2025 04:14:14 GMT"
},
{
"version": "v3",
"created": "Tue, 17 Jun 2025 04:56:46 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Ro",
"Yeonju",
""
],
[
"Zhang",
"Zhenyu",
""
],
[
"Kundu",
"Souvik",
""
],
[
"Wang",
"Zhangyang",
""
],
[
"Akella",
"Aditya",
""
]
] |
2506.09555 | Hyejung Hailey Jee | Hyejung H. Jee, Florian J. Curchod and Mafalda L. Almeida | More entropy from shorter experiments using polytope approximations to the quantum set | 22+9 pages (6 figures) | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | We introduce a systematic method for constructing polytope approximations to the quantum set in a variety of device-independent quantum random number generation (DI-QRNG) protocols. Our approach relies on two general-purpose algorithms that iteratively refine an initial outer-polytope approximation, guided by typical device behaviour and cryptographic intuition. These refinements strike a balance between computational tractability and approximation effectiveness. By integrating these approximations into the probability estimation (PE) framework [Zhang et al., PRA 2018], we obtain significantly improved certified entropy bounds in the finite-size regime. We test our method on various bipartite and tripartite DI-QRNG protocols, using both simulated and experimental data. In all cases, it yields notably higher entropy rates with fewer device uses than the existing techniques. We further extend our analysis to the more demanding task of randomness amplification, demonstrating major performance gains without added complexity. These results offer an effective and ready-to-use method to prove security-with improved certified entropy rates-in the most common practical DI-QRNG protocols. Our algorithms and entropy certification with PE tools are publicly available under a non-commercial license at https://github.com/CQCL/PE_polytope_approximation. | [
{
"version": "v1",
"created": "Wed, 11 Jun 2025 09:37:56 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 12:59:11 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Jee",
"Hyejung H.",
""
],
[
"Curchod",
"Florian J.",
""
],
[
"Almeida",
"Mafalda L.",
""
]
] |
2506.09697 | Paolo Franceschi | Paolo Franceschi, Andrea Bussolan, Vincenzo Pomponi, Oliver Avram, Stefano Baraldo, Anna Valente | Human-robot collaborative transport personalization via Dynamic Movement Primitives and velocity scaling | null | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Nowadays, industries are showing a growing interest in human-robot collaboration, particularly for shared tasks. This requires intelligent strategies to plan a robot's motions, considering both task constraints and human-specific factors such as height and movement preferences. This work introduces a novel approach to generate personalized trajectories using Dynamic Movement Primitives (DMPs), enhanced with real-time velocity scaling based on human feedback. The method was rigorously tested in industrial-grade experiments, focusing on the collaborative transport of an engine cowl lip section. Comparative analysis between DMP-generated trajectories and a state-of-the-art motion planner (BiTRRT) highlights their adaptability combined with velocity scaling. Subjective user feedback further demonstrates a clear preference for DMP- based interactions. Objective evaluations, including physiological measurements from brain and skin activity, reinforce these findings, showcasing the advantages of DMPs in enhancing human-robot interaction and improving user experience. | [
{
"version": "v1",
"created": "Wed, 11 Jun 2025 13:12:56 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 13:22:26 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Franceschi",
"Paolo",
""
],
[
"Bussolan",
"Andrea",
""
],
[
"Pomponi",
"Vincenzo",
""
],
[
"Avram",
"Oliver",
""
],
[
"Baraldo",
"Stefano",
""
],
[
"Valente",
"Anna",
""
]
] |
2506.09934 | Jared Lawson | Jared Lawson and Rohan Chitale and Nabil Simaan | Fluoroscopic Shape and Pose Tracking of Catheters with Custom Radiopaque Markers | 8 pages, 5 figures, accepted in Robotics and Automation Letters | null | 10.1109/LRA.2025.3581043 | null | cs.RO cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Safe navigation of steerable and robotic catheters in the cerebral vasculature requires awareness of the catheters shape and pose. Currently, a significant perception burden is placed on interventionalists to mentally reconstruct and predict catheter motions from biplane fluoroscopy images. Efforts to track these catheters are limited to planar segmentation or bulky sensing instrumentation, which are incompatible with microcatheters used in neurointervention. In this work, a catheter is equipped with custom radiopaque markers arranged to enable simultaneous shape and pose estimation under biplane fluoroscopy. A design measure is proposed to guide the arrangement of these markers to minimize sensitivity to marker tracking uncertainty. This approach was deployed for microcatheters smaller than 2mm OD navigating phantom vasculature with shape tracking errors less than 1mm and catheter roll errors below 40 degrees. This work can enable steerable catheters to autonomously navigate under biplane imaging. | [
{
"version": "v1",
"created": "Wed, 11 Jun 2025 16:56:23 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Lawson",
"Jared",
""
],
[
"Chitale",
"Rohan",
""
],
[
"Simaan",
"Nabil",
""
]
] |
2506.10013 | Joonhyung Bae | Yerin Doh, Joonhyung Bae | Immersive Fantasy Based on Digital Nostalgia: Environmental Narratives for the Korean Millennials and Gen Z | Accepted at ISEA 2025 (International Symposium on Electronic Art) | null | null | null | cs.MM cs.CY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This study introduces the media artwork Dear Passenger, Please Wear a Mask, designed to offer a layered exploration of single-use mask waste, which escalated during the COVID-19 pandemic. The piece reframes underappreciated ecological concerns by interweaving digital nostalgia and airline travel recollections of Millennials and Gen Z with a unique fantasy narrative. Via a point-and-click game and an immersive exhibition, participants traverse both virtual and real domains, facing ethical and environmental dilemmas. While it fosters empathy and potential action, resource use and post-experience engagement challenges persist. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 01:44:03 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 10:10:50 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Doh",
"Yerin",
""
],
[
"Bae",
"Joonhyung",
""
]
] |
2506.10037 | Shawn Zheng Kai Tan | Shawn Zheng Kai Tan, Aleix Puig-Barbe, Damien Goutte-Gattat, Caroline Eastwood, Brian Aevermann, Alida Avola, James P Balhoff, Ismail Ugur Bayindir, Jasmine Belfiore, Anita Reane Caron, David S Fischer, Nancy George, Benjamin M Gyori, Melissa A Haendel, Charles Tapley Hoyt, Huseyin Kir, Tiago Lubiana, Nicolas Matentzoglu, James A Overton, Beverly Peng, Bjoern Peters, Ellen M Quardokus, Patrick L Ray, Paola Roncaglia, Andrea D Rivera, Ray Stefancsik, Wei Kheng Teh, Sabrina Toro, Nicole Vasilevsky, Chuan Xu, Yun Zhang, Richard H Scheuermann, Chirstopher J Mungall, Alexander D Diehl, David Osumi-Sutherland | The Cell Ontology in the age of single-cell omics | 41 pages, 7 Figures | null | null | null | q-bio.OT cs.IR | http://creativecommons.org/licenses/by/4.0/ | Single-cell omics technologies have transformed our understanding of cellular diversity by enabling high-resolution profiling of individual cells. However, the unprecedented scale and heterogeneity of these datasets demand robust frameworks for data integration and annotation. The Cell Ontology (CL) has emerged as a pivotal resource for achieving FAIR (Findable, Accessible, Interoperable, and Reusable) data principles by providing standardized, species-agnostic terms for canonical cell types - forming a core component of a wide range of platforms and tools. In this paper, we describe the wide variety of uses of CL in these platforms and tools and detail ongoing work to improve and extend CL content including the addition of transcriptomically defined types, working closely with major atlasing efforts including the Human Cell Atlas and the Brain Initiative Cell Atlas Network to support their needs. We cover the challenges and future plans for harmonising classical and transcriptomic cell type definitions, integrating markers and using Large Language Models (LLMs) to improve content and efficiency of CL workflows. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 21:38:26 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 13:13:31 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Tan",
"Shawn Zheng Kai",
""
],
[
"Puig-Barbe",
"Aleix",
""
],
[
"Goutte-Gattat",
"Damien",
""
],
[
"Eastwood",
"Caroline",
""
],
[
"Aevermann",
"Brian",
""
],
[
"Avola",
"Alida",
""
],
[
"Balhoff",
"James P",
""
],
[
"Bayindir",
"Ismail Ugur",
""
],
[
"Belfiore",
"Jasmine",
""
],
[
"Caron",
"Anita Reane",
""
],
[
"Fischer",
"David S",
""
],
[
"George",
"Nancy",
""
],
[
"Gyori",
"Benjamin M",
""
],
[
"Haendel",
"Melissa A",
""
],
[
"Hoyt",
"Charles Tapley",
""
],
[
"Kir",
"Huseyin",
""
],
[
"Lubiana",
"Tiago",
""
],
[
"Matentzoglu",
"Nicolas",
""
],
[
"Overton",
"James A",
""
],
[
"Peng",
"Beverly",
""
],
[
"Peters",
"Bjoern",
""
],
[
"Quardokus",
"Ellen M",
""
],
[
"Ray",
"Patrick L",
""
],
[
"Roncaglia",
"Paola",
""
],
[
"Rivera",
"Andrea D",
""
],
[
"Stefancsik",
"Ray",
""
],
[
"Teh",
"Wei Kheng",
""
],
[
"Toro",
"Sabrina",
""
],
[
"Vasilevsky",
"Nicole",
""
],
[
"Xu",
"Chuan",
""
],
[
"Zhang",
"Yun",
""
],
[
"Scheuermann",
"Richard H",
""
],
[
"Mungall",
"Chirstopher J",
""
],
[
"Diehl",
"Alexander D",
""
],
[
"Osumi-Sutherland",
"David",
""
]
] |
2506.10055 | Dingfeng Shi | Dingfeng Shi, Jingyi Cao, Qianben Chen, Weichen Sun, Weizhen Li, Hongxuan Lu, Fangchen Dong, Tianrui Qin, King Zhu, Minghao Liu, Jian Yang, Ge Zhang, Jiaheng Liu, Changwang Zhang, Jun Wang, Yuchen Eleanor Jiang, Wangchunshu Zhou | TaskCraft: Automated Generation of Agentic Tasks | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Agentic tasks, which require multi-step problem solving with autonomy, tool use, and adaptive reasoning, are becoming increasingly central to the advancement of NLP and AI. However, existing instruction data lacks tool interaction, and current agentic benchmarks rely on costly human annotation, limiting their scalability. We introduce \textsc{TaskCraft}, an automated workflow for generating difficulty-scalable, multi-tool, and verifiable agentic tasks with execution trajectories. TaskCraft expands atomic tasks using depth-based and width-based extensions to create structurally and hierarchically complex challenges. Empirical results show that these tasks improve prompt optimization in the generation workflow and enhance supervised fine-tuning of agentic foundation models. We present a large-scale synthetic dataset of approximately 36,000 tasks with varying difficulty to support future research on agent tuning and evaluation. | [
{
"version": "v1",
"created": "Wed, 11 Jun 2025 17:58:14 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 15:19:26 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Shi",
"Dingfeng",
""
],
[
"Cao",
"Jingyi",
""
],
[
"Chen",
"Qianben",
""
],
[
"Sun",
"Weichen",
""
],
[
"Li",
"Weizhen",
""
],
[
"Lu",
"Hongxuan",
""
],
[
"Dong",
"Fangchen",
""
],
[
"Qin",
"Tianrui",
""
],
[
"Zhu",
"King",
""
],
[
"Liu",
"Minghao",
""
],
[
"Yang",
"Jian",
""
],
[
"Zhang",
"Ge",
""
],
[
"Liu",
"Jiaheng",
""
],
[
"Zhang",
"Changwang",
""
],
[
"Wang",
"Jun",
""
],
[
"Jiang",
"Yuchen Eleanor",
""
],
[
"Zhou",
"Wangchunshu",
""
]
] |
2506.10130 | Luciano Floridi | Luciano Floridi | A Conjecture on a Fundamental Trade-Off between Certainty and Scope in Symbolic and Generative AI | version 3 | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | This article introduces a conjecture that formalises a fundamental trade-off between provable correctness and broad data-mapping capacity in Artificial Intelligence (AI) systems. When an AI system is engineered for deductively watertight guarantees (demonstrable certainty about the error-free nature of its outputs) -- as in classical symbolic AI -- its operational domain must be narrowly circumscribed and pre-structured. Conversely, a system that can input high-dimensional data to produce rich information outputs -- as in contemporary generative models -- necessarily relinquishes the possibility of zero-error performance, incurring an irreducible risk of errors or misclassification. By making this previously implicit trade-off explicit and open to rigorous verification, the conjecture significantly reframes both engineering ambitions and philosophical expectations for AI. After reviewing the historical motivations for this tension, the article states the conjecture in information-theoretic form and contextualises it within broader debates in epistemology, formal verification, and the philosophy of technology. It then offers an analysis of its implications and consequences, drawing on notions of underdetermination, prudent epistemic risk, and moral responsibility. The discussion clarifies how, if correct, the conjecture would help reshape evaluation standards, governance frameworks, and hybrid system design. The conclusion underscores the importance of eventually proving or refuting the inequality for the future of trustworthy AI. | [
{
"version": "v1",
"created": "Wed, 11 Jun 2025 19:18:13 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 16:13:48 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Floridi",
"Luciano",
""
]
] |
2506.10232 | Vo Phuc Dang | Dang Vo Phuc | On Singer's conjecture for the fourth algebraic transfer in certain generic degrees | 32 pages. This version serves as a corrigendum to our published paper. It corrects computational inaccuracies in Theorems 2.1.4, 2.2.1, Proposition 4.1.8, Lemata 4.2.1, 4.2.2, 4.2.3 of the original publication. These corrections do not affect the statements or conclusions of the main results, namely Theorems 1.3 and 1.4 | Journal of Homotopy and Related Structures, Vol.19, 431-473 (2024) | 10.1007/s40062-024-00351-8 | null | math.AT math.GT math.RA math.RT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Let $A$ be the Steenrod algebra over the finite field $k := \mathbb Z_2$ and $G(q)$ be the general linear group of rank $q$ over $k.$ A well-known open problem in algebraic topology is the explicit determination of the cohomology groups of the Steenrod algebra, ${\rm Ext}^{q, *}_A(k, k),$ for all homological degrees $q \geq 0.$ The Singer algebraic transfer of rank $q,$ formulated by William Singer in 1989, serves as a valuable method for the description of such Ext groups. This transfer maps from the coinvariants of a certain representation of $G(q)$ to ${\rm Ext}^{q, *}_A(k, k).$ Singer predicted that the algebraic transfer is always injective, but this has gone unanswered for all $q\geq 4.$ This paper establishes Singer's conjecture for rank four in the generic degrees $n = 2^{s+t+1} +2^{s+1} - 3$ whenever $t\neq 3$ and $s\geq 1,$ and $n = 2^{s+t} + 2^{s} - 2$ whenever $t\neq 2,\, 3,\, 4$ and $s\geq 1.$ In conjunction with our previous results, this completes the proof of the Singer conjecture for rank four. | [
{
"version": "v1",
"created": "Wed, 11 Jun 2025 23:20:58 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 14:54:56 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Phuc",
"Dang Vo",
""
]
] |
2506.10274 | Pooneh Mousavi | Pooneh Mousavi, Gallil Maimon, Adel Moumen, Darius Petermann, Jiatong Shi, Haibin Wu, Haici Yang, Anastasia Kuznetsova, Artem Ploujnikov, Ricard Marxer, Bhuvana Ramabhadran, Benjamin Elizalde, Loren Lugosch, Jinyu Li, Cem Subakan, Phil Woodland, Minje Kim, Hung-yi Lee, Shinji Watanabe, Yossi Adi, Mirco Ravanelli | Discrete Audio Tokens: More Than a Survey! | null | null | null | null | cs.SD cs.AI cs.CL eess.AS | http://creativecommons.org/licenses/by/4.0/ | Discrete audio tokens are compact representations that aim to preserve perceptual quality, phonetic content, and speaker characteristics while enabling efficient storage and inference, as well as competitive performance across diverse downstream tasks. They provide a practical alternative to continuous features, enabling the integration of speech and audio into modern large language models (LLMs). As interest in token-based audio processing grows, various tokenization methods have emerged, and several surveys have reviewed the latest progress in the field. However, existing studies often focus on specific domains or tasks and lack a unified comparison across various benchmarks. This paper presents a systematic review and benchmark of discrete audio tokenizers, covering three domains: speech, music, and general audio. We propose a taxonomy of tokenization approaches based on encoder-decoder, quantization techniques, training paradigm, streamability, and application domains. We evaluate tokenizers on multiple benchmarks for reconstruction, downstream performance, and acoustic language modeling, and analyze trade-offs through controlled ablation studies. Our findings highlight key limitations, practical considerations, and open challenges, providing insight and guidance for future research in this rapidly evolving area. For more information, including our main results and tokenizer database, please refer to our website: https://poonehmousavi.github.io/dates-website/. | [
{
"version": "v1",
"created": "Thu, 12 Jun 2025 01:35:43 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 18:38:29 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Mousavi",
"Pooneh",
""
],
[
"Maimon",
"Gallil",
""
],
[
"Moumen",
"Adel",
""
],
[
"Petermann",
"Darius",
""
],
[
"Shi",
"Jiatong",
""
],
[
"Wu",
"Haibin",
""
],
[
"Yang",
"Haici",
""
],
[
"Kuznetsova",
"Anastasia",
""
],
[
"Ploujnikov",
"Artem",
""
],
[
"Marxer",
"Ricard",
""
],
[
"Ramabhadran",
"Bhuvana",
""
],
[
"Elizalde",
"Benjamin",
""
],
[
"Lugosch",
"Loren",
""
],
[
"Li",
"Jinyu",
""
],
[
"Subakan",
"Cem",
""
],
[
"Woodland",
"Phil",
""
],
[
"Kim",
"Minje",
""
],
[
"Lee",
"Hung-yi",
""
],
[
"Watanabe",
"Shinji",
""
],
[
"Adi",
"Yossi",
""
],
[
"Ravanelli",
"Mirco",
""
]
] |
2506.11002 | Justus Bogner | Roberto Verdecchia, Justus Bogner | Notes On Writing Effective Empirical Software Engineering Papers: An Opinionated Primer | Accepted for publication by ACM SIGSOFT Software Engineering Notes (SEN) | null | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While mastered by some, good scientific writing practices within Empirical Software Engineering (ESE) research appear to be seldom discussed and documented. Despite this, these practices are implicit or even explicit evaluation criteria of typical software engineering conferences and journals. In this pragmatic, educational-first document, we want to provide guidance to those who may feel overwhelmed or confused by writing ESE papers, but also those more experienced who still might find an opinionated collection of writing advice useful. The primary audience we had in mind for this paper were our own BSc, MSc, and PhD students, but also students of others. Our documented advice therefore reflects a subjective and personal vision of writing ESE papers. By no means do we claim to be fully objective, generalizable, or representative of the whole discipline. With that being said, writing papers in this way has worked pretty well for us so far. We hope that this guide can at least partially do the same for others. | [
{
"version": "v1",
"created": "Thu, 17 Apr 2025 08:25:01 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 10:45:49 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Verdecchia",
"Roberto",
""
],
[
"Bogner",
"Justus",
""
]
] |
2506.11037 | Aochuan Chen | Aochuan Chen, Yifan Niu, Ziqi Gao, Yujie Sun, Shoujun Liu, Gong Chen, Yang Liu, Jia Li | Mini-Game Lifetime Value Prediction in WeChat | KDD ADS Track 2025 | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | The LifeTime Value (LTV) prediction, which endeavors to forecast the cumulative purchase contribution of a user to a particular item, remains a vital challenge that advertisers are keen to resolve. A precise LTV prediction system enhances the alignment of user interests with meticulously designed advertisements, thereby generating substantial profits for advertisers. Nonetheless, this issue is complicated by the paucity of data typically observed in real-world advertising scenarios. The purchase rate among registered users is often as critically low as 0.1%, resulting in a dataset where the majority of users make only several purchases. Consequently, there is insufficient supervisory signal for effectively training the LTV prediction model. An additional challenge emerges from the interdependencies among tasks with high correlation. It is a common practice to estimate a user's contribution to a game over a specified temporal interval. Varying the lengths of these intervals corresponds to distinct predictive tasks, which are highly correlated. For instance, predictions over a 7-day period are heavily reliant on forecasts made over a 3-day period, where exceptional cases can adversely affect the accuracy of both tasks. In order to comprehensively address the aforementioned challenges, we introduce an innovative framework denoted as Graph-Represented Pareto-Optimal LifeTime Value prediction (GRePO-LTV). Graph representation learning is initially employed to address the issue of data scarcity. Subsequently, Pareto-Optimization is utilized to manage the interdependence of prediction tasks. | [
{
"version": "v1",
"created": "Wed, 21 May 2025 02:28:26 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 09:42:40 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Chen",
"Aochuan",
""
],
[
"Niu",
"Yifan",
""
],
[
"Gao",
"Ziqi",
""
],
[
"Sun",
"Yujie",
""
],
[
"Liu",
"Shoujun",
""
],
[
"Chen",
"Gong",
""
],
[
"Liu",
"Yang",
""
],
[
"Li",
"Jia",
""
]
] |
2506.11054 | Sai Krishna Deepak Kanneganti | Deepak Kanneganti, Sajib Mistry, Sheik Mohammad Mostakim Fattah, Aneesh Krishna, and Monowar Bhuyan | Adaptive Composition of Machine Learning as a Service (MLaaS) for IoT Environments | null | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The dynamic nature of Internet of Things (IoT) environments challenges the long-term effectiveness of Machine Learning as a Service (MLaaS) compositions. The uncertainty and variability of IoT environments lead to fluctuations in data distribution, e.g., concept drift and data heterogeneity, and evolving system requirements, e.g., scalability demands and resource limitations. This paper proposes an adaptive MLaaS composition framework to ensure a seamless, efficient, and scalable MLaaS composition. The framework integrates a service assessment model to identify underperforming MLaaS services and a candidate selection model to filter optimal replacements. An adaptive composition mechanism is developed that incrementally updates MLaaS compositions using a contextual multi-armed bandit optimization strategy. By continuously adapting to evolving IoT constraints, the approach maintains Quality of Service (QoS) while reducing the computational cost associated with recomposition from scratch. Experimental results on a real-world dataset demonstrate the efficiency of our proposed approach. | [
{
"version": "v1",
"created": "Thu, 22 May 2025 11:31:00 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 11:24:33 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Kanneganti",
"Deepak",
""
],
[
"Mistry",
"Sajib",
""
],
[
"Fattah",
"Sheik Mohammad Mostakim",
""
],
[
"Krishna",
"Aneesh",
""
],
[
"Bhuyan",
"Monowar",
""
]
] |
2506.11154 | Sharvari Kamble . | Sharvari Kamble | SLRNet: A Real-Time LSTM-Based Sign Language Recognition System | 9 pages, 5 figures, includes experimental results. Code available at: https://github.com/Khushi-739/SLRNet | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Sign Language Recognition (SLR) plays a crucial role in bridging the communication gap between the hearing-impaired community and society. This paper introduces SLRNet, a real-time webcam-based ASL recognition system using MediaPipe Holistic and Long Short-Term Memory (LSTM) networks. The model processes video streams to recognize both ASL alphabet letters and functional words. With a validation accuracy of 86.7%, SLRNet demonstrates the feasibility of inclusive, hardware-independent gesture recognition. | [
{
"version": "v1",
"created": "Wed, 11 Jun 2025 14:30:47 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Kamble",
"Sharvari",
""
]
] |
2506.11209 | Vinod Grover | Zhengyang Liu, Vinod Grover | A Performance Model for Warp Specialization Kernels | null | null | null | null | cs.PL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents a performance model tailored for warp specialization kernels, focusing on factors such as warp size, tilling size, input matrix size, memory bandwidth, and thread divergence. Our model offers accurate predictions of execution time by leveraging differential equations validated through simulations and experiments. The insights gained from this model not only enhance our understanding of warp specialization techniques but also have practical implications for optimizing GPU-accelerated applications through compiler optimizations, kernel parameter tuning, and algorithm design. | [
{
"version": "v1",
"created": "Thu, 12 Jun 2025 18:16:04 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 18:20:02 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Liu",
"Zhengyang",
""
],
[
"Grover",
"Vinod",
""
]
] |
2506.11347 | Wenqian Ye | Wenqian Ye, Guangtao Zheng, Aidong Zhang | Improving Group Robustness on Spurious Correlation via Evidential Alignment | Accepted at KDD 2025 (Research Track) | null | 10.1145/3711896.3737002 | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Deep neural networks often learn and rely on spurious correlations, i.e., superficial associations between non-causal features and the targets. For instance, an image classifier may identify camels based on the desert backgrounds. While it can yield high overall accuracy during training, it degrades generalization on more diverse scenarios where such correlations do not hold. This problem poses significant challenges for out-of-distribution robustness and trustworthiness. Existing methods typically mitigate this issue by using external group annotations or auxiliary deterministic models to learn unbiased representations. However, such information is costly to obtain, and deterministic models may fail to capture the full spectrum of biases learned by the models. To address these limitations, we propose Evidential Alignment, a novel framework that leverages uncertainty quantification to understand the behavior of the biased models without requiring group annotations. By quantifying the evidence of model prediction with second-order risk minimization and calibrating the biased models with the proposed evidential calibration technique, Evidential Alignment identifies and suppresses spurious correlations while preserving core features. We theoretically justify the effectiveness of our method as capable of learning the patterns of biased models and debiasing the model without requiring any spurious correlation annotations. Empirical results demonstrate that our method significantly improves group robustness across diverse architectures and data modalities, providing a scalable and principled solution to spurious correlations. | [
{
"version": "v1",
"created": "Thu, 12 Jun 2025 22:47:21 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 16:55:20 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Ye",
"Wenqian",
""
],
[
"Zheng",
"Guangtao",
""
],
[
"Zhang",
"Aidong",
""
]
] |
2506.11421 | Jing Dong | Junli Shao, Jing Dong, Dingzhou Wang, Kowei Shih, Dannier Li, Chengrui Zhou | Deep Learning Model Acceleration and Optimization Strategies for Real-Time Recommendation Systems | null | null | null | null | cs.IR cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With the rapid growth of Internet services, recommendation systems play a central role in delivering personalized content. Faced with massive user requests and complex model architectures, the key challenge for real-time recommendation systems is how to reduce inference latency and increase system throughput without sacrificing recommendation quality. This paper addresses the high computational cost and resource bottlenecks of deep learning models in real-time settings by proposing a combined set of modeling- and system-level acceleration and optimization strategies. At the model level, we dramatically reduce parameter counts and compute requirements through lightweight network design, structured pruning, and weight quantization. At the system level, we integrate multiple heterogeneous compute platforms and high-performance inference libraries, and we design elastic inference scheduling and load-balancing mechanisms based on real-time load characteristics. Experiments show that, while maintaining the original recommendation accuracy, our methods cut latency to less than 30% of the baseline and more than double system throughput, offering a practical solution for deploying large-scale online recommendation services. | [
{
"version": "v1",
"created": "Fri, 13 Jun 2025 02:39:21 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 17:08:47 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Shao",
"Junli",
""
],
[
"Dong",
"Jing",
""
],
[
"Wang",
"Dingzhou",
""
],
[
"Shih",
"Kowei",
""
],
[
"Li",
"Dannier",
""
],
[
"Zhou",
"Chengrui",
""
]
] |
2506.11424 | Yoshiko Hayashi | Yoshiko Hayashi | Local empirical Bayes correction for Bayesian modeling | null | Osaka Keidai Ronshu, vol.68, no.4, pp.161-172, 2017 | 10.24644/keidaironshu.68.4_161 | null | stat.ME math.ST stat.TH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The James-Stein estimator has attracted much interest as a shrinkage estimator that yields better estimates than the maximum likelihood estimator. The James-Stein estimator is also very useful as an argument in favor of empirical Bayesian methods. However, for problems involving large-scale data, such as differential gene expression data, the distribution is considered a mixture distribution with different means that cannot be considered sufficiently close. Therefore, it is not appropriate to apply the James-Stein estimator. Efron (2011) proposed a local empirical Bayes correction that attempted to correct a selection bias for large-scale data. | [
{
"version": "v1",
"created": "Fri, 13 Jun 2025 02:43:13 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 00:52:30 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Hayashi",
"Yoshiko",
""
]
] |
2506.11557 | Ying-Jia Lin | Chih-Hao Hsu, Ying-Jia Lin, Hung-Yu Kao | From Persona to Person: Enhancing the Naturalness with Multiple Discourse Relations Graph Learning in Personalized Dialogue Generation | Accepted by PAKDD 2025 | null | 10.1007/978-981-96-8173-0_15 | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In dialogue generation, the naturalness of responses is crucial for effective human-machine interaction. Personalized response generation poses even greater challenges, as the responses must remain coherent and consistent with the user's personal traits or persona descriptions. We propose MUDI ($\textbf{Mu}$ltiple $\textbf{Di}$scourse Relations Graph Learning) for personalized dialogue generation. We utilize a Large Language Model to assist in annotating discourse relations and to transform dialogue data into structured dialogue graphs. Our graph encoder, the proposed DialogueGAT model, then captures implicit discourse relations within this structure, along with persona descriptions. During the personalized response generation phase, novel coherence-aware attention strategies are implemented to enhance the decoder's consideration of discourse relations. Our experiments demonstrate significant improvements in the quality of personalized responses, thus resembling human-like dialogue exchanges. | [
{
"version": "v1",
"created": "Fri, 13 Jun 2025 08:12:52 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Hsu",
"Chih-Hao",
""
],
[
"Lin",
"Ying-Jia",
""
],
[
"Kao",
"Hung-Yu",
""
]
] |
2506.11681 | Pratibha Zunjare | Pratibha Zunjare and Michael Hsiao | A Hybrid Multi-Agent Prompting Approach for Simplifying Complex Sentences | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper addresses the challenge of transforming complex sentences into sequences of logical, simplified sentences while preserving semantic and logical integrity with the help of Large Language Models. We propose a hybrid approach that combines advanced prompting with multi-agent architectures to enhance the sentence simplification process. Experimental results show that our approach was able to successfully simplify 70% of the complex sentences written for video game design application. In comparison, a single-agent approach attained a 48% success rate on the same task. | [
{
"version": "v1",
"created": "Fri, 13 Jun 2025 11:19:27 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 15:59:13 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Zunjare",
"Pratibha",
""
],
[
"Hsiao",
"Michael",
""
]
] |
2506.11750 | Nicola Rizzo | Nicola Rizzo, Manuel C\'aceres, Veli M\"akinen | Practical colinear chaining on sequences revisited | 16 pages, 4 figures, 2 tables, 1 algorithm. To appear in ISBRA2025 (minus the appendix and Table 2) | null | null | null | cs.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Colinear chaining is a classical heuristic for sequence alignment and is widely used in modern practical aligners. Jain et al. (J. Comput. Biol. 2022) proposed an $O(n \log^3 n)$ time algorithm to chain a set of $n$ anchors so that the chaining cost matches the edit distance of the input sequences, when anchors are all the maximal exact matches. Moreover, assuming a uniform and sparse distribution of anchors, they provided a practical solution ($\mathtt{ChainX}$) working in $O(n \cdot \mathrm{SOL} + n \log n)$ average-case time, where $\mathrm{SOL}$ is the cost of the output chain. This practical solution is not guaranteed to be optimal: we study the failing cases, introduce the anchor diagonal distance, and find and implement an optimal algorithm working in $O(n \cdot \mathrm{OPT} + n \log n)$ average-case time, where $\mathrm{OPT}$ $\le \mathrm{SOL}$ is the optimal chaining cost. We validate the results by Jain et al., show that $\mathtt{ChainX}$ can be suboptimal with a realistic long read dataset, and show minimal computational slowdown for our solution. | [
{
"version": "v1",
"created": "Fri, 13 Jun 2025 13:04:04 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 23:11:37 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Rizzo",
"Nicola",
""
],
[
"Cáceres",
"Manuel",
""
],
[
"Mäkinen",
"Veli",
""
]
] |
2506.11964 | Josias Langbehn | Josias Langbehn, George Mouloudakis, Emma King, Rapha\"el Menu, Igor Gornyi, Giovanna Morigi, Yuval Gefen, Christiane P. Koch | Universal cooling of quantum systems via randomized measurements | null | null | null | null | quant-ph cond-mat.mes-hall cond-mat.str-el | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | It is generally believed that the design of cooling protocols requires knowledge of the system's spectrum. In contrast, cooling processes in nature occur whenever the system is coupled to a cold bath. But how does nature know how to cool quantum systems? Here we address this question by mimicking a natural cold bath with a reservoir of "meter" qubits that are initialized in their ground state and interact with the system sequentially for a specified time before being discarded. This protocol is equivalent to quantum measurements without keeping the measurement readouts. We show that a quantum system can be cooled without prior knowledge of the system's details when the interactions between the system and the meters, as well as the level splittings of the meters, are chosen randomly. For sufficiently small interaction strengths and long interaction times, the rotating-wave approximation becomes valid, ensuring that resonant energy-exchange processes, which lead to cooling, dominate over heating. This demonstrates that robust and scalable cooling of complex quantum systems can be achieved through generic, structure-independent protocols. Our findings thus offer a versatile universal framework for engineering, controlling, and manipulating quantum matter far from equilibrium, in particular, for the purposes of quantum information processing and quantum simulations. | [
{
"version": "v1",
"created": "Fri, 13 Jun 2025 17:20:01 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 07:47:21 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Langbehn",
"Josias",
""
],
[
"Mouloudakis",
"George",
""
],
[
"King",
"Emma",
""
],
[
"Menu",
"Raphaël",
""
],
[
"Gornyi",
"Igor",
""
],
[
"Morigi",
"Giovanna",
""
],
[
"Gefen",
"Yuval",
""
],
[
"Koch",
"Christiane P.",
""
]
] |
2506.11990 | Pei-Chun Su | Pei-Chun Su, Ronald R. Coifman | Learning the Analytic Geometry of Transformations to Achieve Efficient Computation | null | null | null | null | math.NA cs.NA | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We propose a novel framework for fast integral operations by uncovering hidden geometries in the row and column structures of the underlying operators. This is accomplished through an iterative procedure that constructs adaptive hierarchical partition trees, revealing latent multiscale organization and exposing local low-rank structures within the data. Guided by this geometry, we employ two complementary techniques: (1) the \emph{butterfly algorithm}, which exploits the learned hierarchical low-rank structure; and (2) \emph{adaptive best tilings} in both space and frequency using all levels of the generalized Haar--Walsh wavelet packet tree. These techniques enable efficient matrix factorization and multiplication. Unlike classical approaches that rely on prior knowledge of the underlying geometry, our method is fully data-driven and applicable to matrices arising from irregular or unknown distributions. We demonstrate the effectiveness of our approach on matrices associated with acoustic heterogeneous potential operators and families of orthogonal polynomials. The resulting compressed representations reduce storage complexity from $\mathcal{O}(N^2)$ to $\mathcal{O}(N \log N)$, enabling fast computation and scalable implementation. | [
{
"version": "v1",
"created": "Fri, 13 Jun 2025 17:47:32 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2025 19:16:15 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Su",
"Pei-Chun",
""
],
[
"Coifman",
"Ronald R.",
""
]
] |
2506.12308 | Weijie Yuan | Weijie Yuan, Yuanhao Cui, Jiacheng Wang, Fan Liu, Geng Sun, Tao Xiang, Jie Xu, Shi Jin, Dusit Niyato, Sinem Coleri, Sumei Sun, Shiwen Mao, Abbas Jamalipour, Dong In Kim, Mohamed-Slim Alouini, and Xuemin Shen | From Ground to Sky: Architectures, Applications, and Challenges Shaping Low-Altitude Wireless Networks | 10 pages, 5 figures | null | null | null | eess.SP cs.SY eess.SY | http://creativecommons.org/licenses/by/4.0/ | In this article, we introduce a novel low-altitude wireless network (LAWN), which is a reconfigurable, three-dimensional (3D) layered architecture. In particular, the LAWN integrates connectivity, sensing, control, and computing across aerial and terrestrial nodes that enable seamless operation in complex, dynamic, and mission-critical environments. Different from the conventional aerial communication systems, LAWN's distinctive feature is its tight integration of functional planes in which multiple functionalities continually reshape themselves to operate safely and efficiently in the low-altitude sky. With the LAWN, we discuss several enabling technologies, such as integrated sensing and communication (ISAC), semantic communication, and fully-actuated control systems. Finally, we identify potential applications and key cross-layer challenges. This article offers a comprehensive roadmap for future research and development in the low-altitude airspace. | [
{
"version": "v1",
"created": "Sat, 14 Jun 2025 02:04:41 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 01:15:55 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Yuan",
"Weijie",
""
],
[
"Cui",
"Yuanhao",
""
],
[
"Wang",
"Jiacheng",
""
],
[
"Liu",
"Fan",
""
],
[
"Sun",
"Geng",
""
],
[
"Xiang",
"Tao",
""
],
[
"Xu",
"Jie",
""
],
[
"Jin",
"Shi",
""
],
[
"Niyato",
"Dusit",
""
],
[
"Coleri",
"Sinem",
""
],
[
"Sun",
"Sumei",
""
],
[
"Mao",
"Shiwen",
""
],
[
"Jamalipour",
"Abbas",
""
],
[
"Kim",
"Dong In",
""
],
[
"Alouini",
"Mohamed-Slim",
""
],
[
"Shen",
"Xuemin",
""
]
] |
2506.12340 | Zongyu Wu | Zongyu Wu, Minhua Lin, Zhiwei Zhang, Fali Wang, Xianren Zhang, Xiang Zhang, Suhang Wang | Image Corruption-Inspired Membership Inference Attacks against Large Vision-Language Models | Preprint. 15 pages | null | null | null | cs.CV cs.CR | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Large vision-language models (LVLMs) have demonstrated outstanding performance in many downstream tasks. However, LVLMs are trained on large-scale datasets, which can pose privacy risks if training images contain sensitive information. Therefore, it is important to detect whether an image is used to train the LVLM. Recent studies have investigated membership inference attacks (MIAs) against LVLMs, including detecting image-text pairs and single-modality content. In this work, we focus on detecting whether a target image is used to train the target LVLM. We design simple yet effective Image Corruption-Inspired Membership Inference Attacks (ICIMIA) against LLVLMs, which are inspired by LVLM's different sensitivity to image corruption for member and non-member images. We first perform an MIA method under the white-box setting, where we can obtain the embeddings of the image through the vision part of the target LVLM. The attacks are based on the embedding similarity between the image and its corrupted version. We further explore a more practical scenario where we have no knowledge about target LVLMs and we can only query the target LVLMs with an image and a question. We then conduct the attack by utilizing the output text embeddings' similarity. Experiments on existing datasets validate the effectiveness of our proposed attack methods under those two different settings. | [
{
"version": "v1",
"created": "Sat, 14 Jun 2025 04:22:36 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 04:52:16 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Wu",
"Zongyu",
""
],
[
"Lin",
"Minhua",
""
],
[
"Zhang",
"Zhiwei",
""
],
[
"Wang",
"Fali",
""
],
[
"Zhang",
"Xianren",
""
],
[
"Zhang",
"Xiang",
""
],
[
"Wang",
"Suhang",
""
]
] |
2506.12347 | Aayush Kumar | Aayush Kumar, Yasharth Bajpai, Sumit Gulwani, Gustavo Soares, Emerson Murphy-Hill | Sharp Tools: How Developers Wield Agentic AI in Real Software Engineering Tasks | null | null | null | null | cs.SE cs.HC | http://creativecommons.org/licenses/by/4.0/ | Software Engineering Agents (SWE agents) can autonomously perform development tasks on benchmarks like SWE Bench, but still face challenges when tackling complex and ambiguous real-world tasks. Consequently, SWE agents are often designed to allow interactivity with developers, enabling collaborative problem-solving. To understand how developers collaborate with SWE agents and the communication challenges that arise in such interactions, we observed 19 developers using an in-IDE agent to resolve 33 open issues in repositories to which they had previously contributed. Participants successfully resolved about half of these issues, with participants solving issues incrementally having greater success than those using a one-shot approach. Participants who actively collaborated with the agent and iterated on its outputs were also more successful, though they faced challenges in trusting the agent's responses and collaborating on debugging and testing. These results have implications for successful developer-agent collaborations, and for the design of more effective SWE agents. | [
{
"version": "v1",
"created": "Sat, 14 Jun 2025 04:52:01 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 17:49:35 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Kumar",
"Aayush",
""
],
[
"Bajpai",
"Yasharth",
""
],
[
"Gulwani",
"Sumit",
""
],
[
"Soares",
"Gustavo",
""
],
[
"Murphy-Hill",
"Emerson",
""
]
] |
2506.12376 | Zhaochen Hong | Zhaochen Hong, Haofei Yu, Jiaxuan You | ConsistencyChecker: Tree-based Evaluation of LLM Generalization Capabilities | Accepted at ACL 2025 Main Conference | null | null | null | cs.AI cs.CL | http://creativecommons.org/licenses/by/4.0/ | Evaluating consistency in large language models (LLMs) is crucial for ensuring reliability, particularly in complex, multi-step interactions between humans and LLMs. Traditional self-consistency methods often miss subtle semantic changes in natural language and functional shifts in code or equations, which can accumulate over multiple transformations. To address this, we propose ConsistencyChecker, a tree-based evaluation framework designed to measure consistency through sequences of reversible transformations, including machine translation tasks and AI-assisted programming tasks. In our framework, nodes represent distinct text states, while edges correspond to pairs of inverse operations. Dynamic and LLM-generated benchmarks ensure a fair assessment of the model's generalization ability and eliminate benchmark leakage. Consistency is quantified based on similarity across different depths of the transformation tree. Experiments on eight models from various families and sizes show that ConsistencyChecker can distinguish the performance of different models. Notably, our consistency scores-computed entirely without using WMT paired data-correlate strongly (r > 0.7) with WMT 2024 auto-ranking, demonstrating the validity of our benchmark-free approach. Our implementation is available at: https://github.com/ulab-uiuc/consistencychecker. | [
{
"version": "v1",
"created": "Sat, 14 Jun 2025 07:18:33 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 08:11:59 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Hong",
"Zhaochen",
""
],
[
"Yu",
"Haofei",
""
],
[
"You",
"Jiaxuan",
""
]
] |
2506.12380 | Ipsita Mandal | Ipsita Mandal | Distinguishing features of longitudinal magnetoconductivity for a Rarita-Schwinger-Weyl node | we have reviewed the Boltzmann formalism for the ease of readability, although the framework is the same as used in arXiv:2505.19636 and arXiv:2506.07913 | null | null | null | cond-mat.mes-hall cond-mat.str-el hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The band-degeneracy points in the Brillouin zones of chiral crystals exist in multiple avatars, with the high-symmetry points being able to host multifold nodes of distinct characters. A class of such crystals, assisted by the spin-orbit coupling, harbours fourfold degeneracy in the form of Rarita-Schwinger-Weyl node (RSWN) at the $\Gamma$-point. Our aim is to explore the nature of longitudinal magnetoconductivity, arising from applying collinear electric and magnetic fields, for such systems. Adjusting the chemical potential to lie near the intrinsic energy-location of the RSWN, the multifold nature of the RSWN is revealed by an interplay of intraband and interband scatterings, which would not arise in twofold degeneracies like the conventional Weyl nodes. The current study fills up the much-needed gap in obtaining the linear response from an exact computation, rather than the insufficient relaxation-time approximation employed earlier. | [
{
"version": "v1",
"created": "Sat, 14 Jun 2025 07:29:55 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Mandal",
"Ipsita",
""
]
] |
2506.12386 | Iannis Kominis | I. K. Kominis and C. Kosmas | Prospecting for lunar Helium-3 with a radio-frequency atomic magnetometer | 7 pages, 1 figure | null | null | null | physics.atom-ph physics.app-ph physics.ins-det physics.space-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Mining $^3{\rm He}$ from lunar regolith has attracted significant interest in recent years due to the scarcity of $^3{\rm He}$ on Earth and its diverse applications, from cryogenics and medical imaging, to nuclear physics and future nuclear fusion. Given the stringent technical and economic challenges of mining lunar $^3{\rm He}$, precise prospecting is essential. Here we propose a prospecting methodology based on a radio-frequency atomic magnetometer, which can detect the dipolar magnetic field of thermally polarized $^3{\rm He}$ spins. With a 200 g regolith sample and an rf-magnetometer with sensitivity $1~{\rm fT/\sqrt{Hz}}$ we can detect $^3{\rm He}$ with abundance 5 ppb within a measurement time of just 5 min. The associated apparatus is lightweight and significantly more cost-effective than alternative measurement techniques. The proposed prospecting method is readily deployable and could substantially improve the technical and economic feasibility of mining lunar $^3{\rm He}$. | [
{
"version": "v1",
"created": "Sat, 14 Jun 2025 07:48:09 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 04:52:31 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Kominis",
"I. K.",
""
],
[
"Kosmas",
"C.",
""
]
] |
2506.12417 | Martijn de Vos | Zachary Doucet, Rishi Sharma, Martijn de Vos, Rafael Pires, Anne-Marie Kermarrec, Oana Balmau | HarMoEny: Efficient Multi-GPU Inference of MoE Models | null | null | null | null | cs.DC | http://creativecommons.org/licenses/by/4.0/ | Mixture-of-Experts (MoE) models offer computational efficiency during inference by activating only a subset of specialized experts for a given input. This enables efficient model scaling on multi-GPU systems that use expert parallelism without compromising performance. However, load imbalance among experts and GPUs introduces waiting times, which can significantly increase inference latency. To address this challenge, we propose HarMoEny, a novel solution to address MoE load imbalance through two simple techniques: (i) dynamic token redistribution to underutilized GPUs and (ii) asynchronous prefetching of experts from the system to GPU memory. These techniques achieve a near-perfect load balance among experts and GPUs and mitigate delays caused by overloaded GPUs. We implement HarMoEny and compare its latency and throughput with four MoE baselines using real-world and synthetic datasets. Under heavy load imbalance, HarMoEny increases throughput by 37%-70% and reduces time-to-first-token by 34%-41%, compared to the next-best baseline. Moreover, our ablation study demonstrates that HarMoEny's scheduling policy reduces the GPU idling time by up to 84% compared to the baseline policies. | [
{
"version": "v1",
"created": "Sat, 14 Jun 2025 09:21:52 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 08:25:12 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Doucet",
"Zachary",
""
],
[
"Sharma",
"Rishi",
""
],
[
"de Vos",
"Martijn",
""
],
[
"Pires",
"Rafael",
""
],
[
"Kermarrec",
"Anne-Marie",
""
],
[
"Balmau",
"Oana",
""
]
] |
2506.12468 | Suyeon Kim | Suyeon Kim, SeongKu Kang, Dongwoo Kim, Jungseul Ok, Hwanjo Yu | Delving into Instance-Dependent Label Noise in Graph Data: A Comprehensive Study and Benchmark | 12 pages | KDD 2025 | 10.1145/3711896.3711896.3737376 | null | cs.LG cs.AI stat.ML | http://creativecommons.org/licenses/by/4.0/ | Graph Neural Networks (GNNs) have achieved state-of-the-art performance in node classification tasks but struggle with label noise in real-world data. Existing studies on graph learning with label noise commonly rely on class-dependent label noise, overlooking the complexities of instance-dependent noise and falling short of capturing real-world corruption patterns. We introduce BeGIN (Benchmarking for Graphs with Instance-dependent Noise), a new benchmark that provides realistic graph datasets with various noise types and comprehensively evaluates noise-handling strategies across GNN architectures, noisy label detection, and noise-robust learning. To simulate instance-dependent corruptions, BeGIN introduces algorithmic methods and LLM-based simulations. Our experiments reveal the challenges of instance-dependent noise, particularly LLM-based corruption, and underscore the importance of node-specific parameterization to enhance GNN robustness. By comprehensively evaluating noise-handling strategies, BeGIN provides insights into their effectiveness, efficiency, and key performance factors. We expect that BeGIN will serve as a valuable resource for advancing research on label noise in graphs and fostering the development of robust GNN training methods. The code is available at https://github.com/kimsu55/BeGIN. | [
{
"version": "v1",
"created": "Sat, 14 Jun 2025 12:14:15 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 03:17:11 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Kim",
"Suyeon",
""
],
[
"Kang",
"SeongKu",
""
],
[
"Kim",
"Dongwoo",
""
],
[
"Ok",
"Jungseul",
""
],
[
"Yu",
"Hwanjo",
""
]
] |
2506.12508 | Wentao Zhang | Wentao Zhang, Ce Cui, Yilei Zhao, Rui Hu, Yang Liu, Yahui Zhou, Bo An | AgentOrchestra: A Hierarchical Multi-Agent Framework for General-Purpose Task Solving | null | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | Recent advances in agent systems based on large language models (LLMs) have demonstrated strong capabilities in solving complex tasks. However, most current methods lack mechanisms for coordinating specialized agents and have limited ability to generalize to new or diverse domains. We introduce \projectname, a hierarchical multi-agent framework for general-purpose task solving that integrates high-level planning with modular agent collaboration. Inspired by the way a conductor orchestrates a symphony and guided by the principles of \textit{extensibility}, \textit{multimodality}, \textit{modularity}, and \textit{coordination}, \projectname features a central planning agent that decomposes complex objectives and delegates sub-tasks to a team of specialized agents. Each sub-agent is equipped with general programming and analytical tools, as well as abilities to tackle a wide range of real-world specific tasks, including data analysis, file operations, web navigation, and interactive reasoning in dynamic multimodal environments. \projectname supports flexible orchestration through explicit sub-goal formulation, inter-agent communication, and adaptive role allocation. We evaluate the framework on three widely used benchmark datasets covering various real-world tasks, searching web pages, reasoning over heterogeneous modalities, etc. Experimental results demonstrate that \projectname consistently outperforms flat-agent and monolithic baselines in task success rate and adaptability. These findings highlight the effectiveness of hierarchical organization and role specialization in building scalable and general-purpose LLM-based agent systems. | [
{
"version": "v1",
"created": "Sat, 14 Jun 2025 13:45:37 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 07:08:27 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Zhang",
"Wentao",
""
],
[
"Cui",
"Ce",
""
],
[
"Zhao",
"Yilei",
""
],
[
"Hu",
"Rui",
""
],
[
"Liu",
"Yang",
""
],
[
"Zhou",
"Yahui",
""
],
[
"An",
"Bo",
""
]
] |
2506.12526 | Anindya Das | Ujjal Roy, Sourav Manna, Souvik Chakraborty, Kenji Watanabe, Takashi Taniguchi, Ankur Das, Moshe Goldstein, Yuval Gefen and Anindya Das | Half-integer thermal conductance in the absence of Majorana mode | null | null | null | null | cond-mat.mes-hall cond-mat.str-el | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Considering a range of candidate quantum phases of matter, half-integer thermal conductance ($\kappa_{\text{th}}$) is believed to be an unambiguous evidence of non-Abelian states. It has been long known that such half-integer values arise due to the presence of Majorana edge modes, representing a significant step towards topological quantum computing platforms. Here we break this long-standing paradigm, reporting a comprehensive theoretical and experimental study where half-integer two-terminal thermal conductance plateau is realized employing Abelian phases. Our proposed setup features a confined geometry of bilayer graphene, interfacing distinct particle-like and hole-like integer quantum Hall states. Each segment of the device exhibits full charge and thermal equilibration. Our approach is amenable to generalization to other quantum Hall platforms, and may give rise to other values of fractional (electrical and thermal) quantized transport. Our study demonstrates that the observation of robust non-integer values of thermal conductance can arise as a manifestation of mundane equilibration dynamics as opposed to underlying non-trivial topology. | [
{
"version": "v1",
"created": "Sat, 14 Jun 2025 14:52:50 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Roy",
"Ujjal",
""
],
[
"Manna",
"Sourav",
""
],
[
"Chakraborty",
"Souvik",
""
],
[
"Watanabe",
"Kenji",
""
],
[
"Taniguchi",
"Takashi",
""
],
[
"Das",
"Ankur",
""
],
[
"Goldstein",
"Moshe",
""
],
[
"Gefen",
"Yuval",
""
],
[
"Das",
"Anindya",
""
]
] |
2506.12572 | Giovanni Giarda | G. Giarda, A. I. Renzini, C. Pacilio, D. Gerosa | Accelerated inference of binary black-hole populations from the stochastic gravitational-wave background | null | null | null | null | gr-qc astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Third-generation ground-based gravitational wave detectors are expected to observe $\mathcal{O}(10^5)$ of overlapping signals per year from a multitude of astrophysical sources that will be computationally challenging to resolve individually. On the other hand, the stochastic background resulting from the entire population of sources encodes information about the underlying population, allowing for population parameter inference independent and complementary to that obtained with individually resolved events. Parameter estimation in this case is still computationally challenging, as computing the power spectrum involves sampling $\sim 10^5$ sources for each set of hyperparameters describing the binary population. In this work, we build on recently developed importance sampling techniques to compute the SGWB efficiently and train neural networks to interpolate the resulting background. We show that a multi-layer perceptron can encode the model information, allowing for significantly faster inference. We test the network assuming an observing setup with CE and ET sensitivities, where for the first time we include the intrinsic variance of the SGWB in the inference, as in this setup it presents a dominant source of measurement noise. | [
{
"version": "v1",
"created": "Sat, 14 Jun 2025 16:57:36 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Giarda",
"G.",
""
],
[
"Renzini",
"A. I.",
""
],
[
"Pacilio",
"C.",
""
],
[
"Gerosa",
"D.",
""
]
] |
2506.12602 | Yu-Xuan Yin | Yu-xuan Yin, En-kun Li, Bing Zhang and Yi-Ming Hu | Exploring the Link between Fast Radio Burst and Binary Neutron Star Origins with Spaceborne Gravitational Wave Observations | The Astrophysical Journal Letters, Volume 985, Number 2 | null | 10.3847/2041-8213/add875 | null | astro-ph.HE astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | The origin of repeating Fast Radio Bursts (FRBs) is an open question, with observations suggesting that at least some are associated with old stellar populations. It has been proposed that some repeating FRBs may be produced by interactions of the binary neutron star magnetospheres decades to centuries before the coalescence. These systems would also emit centi-Hertz gravitational waves during this period, which can be detectable by space-borne gravitational wave detectors. We explore the prospects of using current and future space-borne gravitational wave detectors, such as TianQin, LISA, and DECIGO, to test this FRB formation hypothesis. Focusing on nearby galaxies like M81, which hosts a repeating FRB source in a globular cluster, we calculate the detection capabilities for binary neutron star systems. Our analysis reveals that while missions like TianQin and LISA face limitations in horizon distance, changing detector pointing direction could significantly enhance detection probabilities. Considering the chance of a Milky Way-like galaxy coincidentally containing a BNS within 100 years before merger is only $3\times10^{-5}$ to $5\times10^{-3}$, if a signal is detected originating from M81, we can establish the link between FRB and binary neutron stars with a significance level of at least 2.81$\sigma$, or a Bayes factor of $4\times10^6 - 7\times10^8$ / $5\times10^2 - 10^5$ against the background model with optimistic/realistic assumptions. Next-generation detectors such as DECIGO offer enhanced capabilities and should easily detect these systems in M81 and beyond. Our work highlights the critical role of space-borne gravitational wave missions in unraveling FRB origins. | [
{
"version": "v1",
"created": "Sat, 14 Jun 2025 18:48:47 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 06:55:51 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Yin",
"Yu-xuan",
""
],
[
"Li",
"En-kun",
""
],
[
"Zhang",
"Bing",
""
],
[
"Hu",
"Yi-Ming",
""
]
] |
2506.12605 | Yutong Zhang | Yutong Zhang, Dora Zhao, Jeffrey T. Hancock, Robert Kraut, Diyi Yang | The Rise of AI Companions: How Human-Chatbot Relationships Influence Well-Being | null | null | null | null | cs.HC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | As large language models (LLMs)-enhanced chatbots grow increasingly expressive and socially responsive, many users are beginning to form companionship-like bonds with them, particularly with simulated AI partners designed to mimic emotionally attuned interlocutors. These emerging AI companions raise critical questions: Can such systems fulfill social needs typically met by human relationships? How do they shape psychological well-being? And what new risks arise as users develop emotional ties to non-human agents? This study investigates how people interact with AI companions, especially simulated partners on CharacterAI, and how this use is associated with users' psychological well-being. We analyzed survey data from 1,131 users and 4,363 chat sessions (413,509 messages) donated by 244 participants, focusing on three dimensions of use: nature of the interaction, interaction intensity, and self-disclosure. By triangulating self-reports primary motivation, open-ended relationship descriptions, and annotated chat transcripts, we identify patterns in how users engage with AI companions and its associations with well-being. Findings suggest that people with smaller social networks are more likely to turn to chatbots for companionship, but that companionship-oriented chatbot usage is consistently associated with lower well-being, particularly when people use the chatbots more intensively, engage in higher levels of self-disclosure, and lack strong human social support. Even though some people turn to chatbots to fulfill social needs, these uses of chatbots do not fully substitute for human connection. As a result, the psychological benefits may be limited, and the relationship could pose risks for more socially isolated or emotionally vulnerable users. | [
{
"version": "v1",
"created": "Sat, 14 Jun 2025 19:00:37 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Jun 2025 08:38:29 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Zhang",
"Yutong",
""
],
[
"Zhao",
"Dora",
""
],
[
"Hancock",
"Jeffrey T.",
""
],
[
"Kraut",
"Robert",
""
],
[
"Yang",
"Diyi",
""
]
] |
2506.12647 | El Arbi Belfarsi | El Arbi Belfarsi, Sophie Brubaker, Maria Valero | Optimizing Blood Transfusions and Predicting Shortages in Resource-Constrained Areas | 12 pages, 9 figures, International Conference on Health Informatics | Proceedings of the 18th International Joint Conference on Biomedical Engineering Systems and Technologies 2025 | 10.5220/0013182700003911 | null | cs.AI cs.LG | http://creativecommons.org/licenses/by-sa/4.0/ | Our research addresses the critical challenge of managing blood transfusions and optimizing allocation in resource-constrained regions. We present heuristic matching algorithms for donor-patient and blood bank selection, alongside machine learning methods to analyze blood transfusion acceptance data and predict potential shortages. We developed simulations to optimize blood bank operations, progressing from random allocation to a system incorporating proximity-based selection, blood type compatibility, expiration prioritization, and rarity scores. Moving from blind matching to a heuristic-based approach yielded a 28.6% marginal improvement in blood request acceptance, while a multi-level heuristic matching resulted in a 47.6% improvement. For shortage prediction, we compared Long Short-Term Memory (LSTM) networks, Linear Regression, and AutoRegressive Integrated Moving Average (ARIMA) models, trained on 170 days of historical data. Linear Regression slightly outperformed others with a 1.40% average absolute percentage difference in predictions. Our solution leverages a Cassandra NoSQL database, integrating heuristic optimization and shortage prediction to proactively manage blood resources. This scalable approach, designed for resource-constrained environments, considers factors such as proximity, blood type compatibility, inventory expiration, and rarity. Future developments will incorporate real-world data and additional variables to improve prediction accuracy and optimization performance. | [
{
"version": "v1",
"created": "Sat, 14 Jun 2025 22:17:45 GMT"
}
] | 2025-06-18T00:00:00 | [
[
"Belfarsi",
"El Arbi",
""
],
[
"Brubaker",
"Sophie",
""
],
[
"Valero",
"Maria",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.