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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2505.08463 | Fujun Zhang | Fujun Zhang and Xiaoying Fan and XiangDong Su and Guanglai Gao | RepCali: High Efficient Fine-tuning Via Representation Calibration in Latent Space for Pre-trained Language Models | 13 pages, 4 figures | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Fine-tuning pre-trained language models (PLMs) has become a dominant paradigm in applying PLMs to downstream tasks. However, with limited fine-tuning, PLMs still struggle with the discrepancies between the representation obtained from the PLMs' encoder and the optimal input to the PLMs' decoder. This paper tackles this challenge by learning to calibrate the representation of PLMs in the latent space. In the proposed representation calibration method (RepCali), we integrate a specific calibration block to the latent space after the encoder and use the calibrated output as the decoder input. The merits of the proposed RepCali include its universality to all PLMs with encoder-decoder architectures, its plug-and-play nature, and ease of implementation. Extensive experiments on 25 PLM-based models across 8 tasks (including both English and Chinese datasets) demonstrate that the proposed RepCali offers desirable enhancements to PLMs (including LLMs) and significantly improves the performance of downstream tasks. Comparison experiments across 4 benchmark tasks indicate that RepCali is superior to the representative fine-tuning baselines. | [
{
"version": "v1",
"created": "Tue, 13 May 2025 11:47:00 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 05:01:48 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhang",
"Fujun",
""
],
[
"Fan",
"Xiaoying",
""
],
[
"Su",
"XiangDong",
""
],
[
"Gao",
"Guanglai",
""
]
] |
2505.08944 | Shaoyu Wang | Shaoyu Wang, Guangrong He, Geon-Woo Kim, Yanqi Zhou, Seo Jin Park | Toward Cost-Efficient Serving of Mixture-of-Experts with Asynchrony | null | null | null | null | cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Mixture-of-Experts (MoE) architectures offer the promise of larger model capacity without the prohibitive costs of fully dense designs. However, in real-world inference serving, load skew across experts often leads to suboptimal device utilization and excessive synchronization overheads. This paper introduces Asynchronous Expert Parallelism (AEP), a new paradigm that decouples layer execution from barrier-style synchronization. By dynamically queuing tokens at each layer (referred to as $\mu$-queuing) and adaptively re-batching them on demand, GPUs avoid waiting for straggling experts and instead continuously process whichever layer is ready. This asynchronous approach mitigates two major inefficiencies in traditional expert-parallel systems: (1) idle GPU time while waiting for the hottest expert, and (2) small-batch executions on colder experts that waste memory bandwidth.
We implement these ideas in a serving system called AMoE, which disaggregates attention from expert layers and uses a defragging scheduler to reduce batch fragmentation. Evaluations on prototype MoE models show that AMoE improves throughput by up to 2.7x compared to state-of-the-art baselines, incurring a manageable latency penalty and providing a cost-effective operating point. Furthermore, experiments demonstrate nearly linear scalability to multi-node settings, whereas the baseline system shows no throughput increase even when the number of GPUs is doubled. | [
{
"version": "v1",
"created": "Tue, 13 May 2025 20:19:23 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 05:20:04 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wang",
"Shaoyu",
""
],
[
"He",
"Guangrong",
""
],
[
"Kim",
"Geon-Woo",
""
],
[
"Zhou",
"Yanqi",
""
],
[
"Park",
"Seo Jin",
""
]
] |
2505.09084 | Shuai Dong | Ziwen Wang, Shuai Dong | Polarization switching in sliding ferroelectrics: the roles of fluctuation and domain wall | 6 pages, 4 figures | Physical Review B 111, L201406 (2025) | 10.1103/PhysRevB.111.L201406 | null | cond-mat.mtrl-sci cond-mat.mes-hall | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Sliding ferroelectricity is highly attractive for its low energy barriers and fatigue resistance. As the origin of these exotic properties, its unconventional switching dynamics remains poorly understood: how an electric field drives a perpendicular sliding? Taking $h$-BN bilayer as a model system, its switching dynamics is studied using \textit{ab initio} calculations. The off-diagonal Born effective charge leads to the perpendicular relationship between the electric field and ionic movements. Interestingly, the rules of intrinsic coercive field are distinct between $h$-BN bilayer and conventional ferroelectrics. For $h$-BN bilayer, any perturbation breaking the in-plane symmetry plays a key role to assist the avalanche-like switching dynamics. Moreover, the exotic large off-diagonal Born effective charge near the $P=0$ intermediate state results in a wriggling motion of domain walls in $h$-BN bilayer. Our results reveal the key factors in the ferroelectric switching of sliding ferroelectrics at room temperature. | [
{
"version": "v1",
"created": "Wed, 14 May 2025 02:36:40 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wang",
"Ziwen",
""
],
[
"Dong",
"Shuai",
""
]
] |
2505.09280 | Tianjun Li | Junle Pei, Yaquan Fang, Lina Wu, Da Xu, Mustapha Biyabi, and Tianjun Li | Quantum Entanglement Theory and Its Generic Searches in High Energy Physics | 42 pages, 6 figures | null | null | null | hep-ph hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose a new formalism for quantum entanglement (QE), and study its generic searches at the colliders. For a general quantum system with $N$ particles, we show that the quantum space (the total spin polarization parameter space) is complex projective space, and the classical space (the spin polarization parameter space for classical theory) is the cartesian product of the complex projective spaces. Thus, the quantum entanglement space is the difference of these two spaces. For the $ff$, $AA$, $Af$, $fff$, and $ffA$ systems, we propose their discriminants $\Delta_i$. The corresponding classical spaces are the discriminant locus $\Delta=0$ for $ff$ system, and intersections of the discriminant loci $\Delta_i=0$ for $AA$, $Af$, $fff$, and $ffA$ systems in the quantum space. In particular, for two fermion $ff$ system, we prove that our discriminant criterion is equivalent to the original Peres-Horodecki criterion and the CHSH criterion. And thus our quantum entanglement space is indeed Bell non-local. With the collider searches, we can reconstruct the discriminants from various measurements, and probe the quantum entanglement spaces via a fundamental approach at exact level. In addition, for the specific approach, we present a comprehensive framework to detect quantum entanglement in high-energy multi-particle systems, spanning fermion pairs ($t\bar{t}$, $\tau^{+}\tau^{-}$), bosonic pairs ($W^{-}W^{+}$), and hybrid or three-body systems ($W^{-}t$, $ttt$, $t\bar{t}W^{-}$), by diverse observables through angular correlations in decay products. These results establish model-independent methodologies for probing QE across collider experiments, bridging quantum information principles with high-energy phenomenology, while offering novel pathways to explore exotic particles and quantum properties in multi-particle systems. | [
{
"version": "v1",
"created": "Wed, 14 May 2025 11:00:43 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 06:51:53 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Pei",
"Junle",
""
],
[
"Fang",
"Yaquan",
""
],
[
"Wu",
"Lina",
""
],
[
"Xu",
"Da",
""
],
[
"Biyabi",
"Mustapha",
""
],
[
"Li",
"Tianjun",
""
]
] |
2505.09335 | Jianjun Jin | Jianjun Jin | Boundedness and norm of certain p-adic Hardy-Littlewood-P\'{o}lya-type operators | 22 pages | null | null | null | math.FA | http://creativecommons.org/licenses/by/4.0/ | In this paper, by introducing some parameters, we define and study certain $p$-adic Hardy-Littlewood-P\'{o}lya-type integral operators acting on $p$-adic weighted Lebesgue spaces. We completely characterize the boundedness of these operators in terms of the parameters. For some special cases, we obtain sharp norm estimates for the operators. These results are not only a complement to some previous results but also an extension of existing ones in the literature. | [
{
"version": "v1",
"created": "Wed, 14 May 2025 12:30:50 GMT"
},
{
"version": "v2",
"created": "Thu, 15 May 2025 03:45:07 GMT"
},
{
"version": "v3",
"created": "Sun, 18 May 2025 06:53:21 GMT"
},
{
"version": "v4",
"created": "Thu, 29 May 2025 01:02:00 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Jin",
"Jianjun",
""
]
] |
2505.09409 | Nam Le | Nam Q. Le | Large dimension behavior of the Hessian eigenvalues of the unit balls | v2: added Remark 1.6 and a reference | null | null | null | math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that a sequence of $k$-Hessian eigenvalues of the unit ball in ${\mathbb R}^n$ stays bounded as long as the ratio $n/k$ stays bounded. Moreover, we identify their growth of order at least $(2-1/k)$ in $n/k$. In the case $k=n$, we show that the Monge--Amp\`ere eigenvalues of the unit balls tend to $4$ in the large dimension limit. | [
{
"version": "v1",
"created": "Wed, 14 May 2025 14:05:33 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 14:39:20 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Le",
"Nam Q.",
""
]
] |
2505.09543 | Karnik Ram | Karnik Ram, Jacobus Dijkman, Ren\'e van Roij, Jan-Willem van de Meent, Bernd Ensing, Max Welling, Daniel Cremers | Learned Free-Energy Functionals from Pair-Correlation Matching for Dynamical Density Functional Theory | 10 pages, 6 figures (see http://karnikram.info/neural-ddft-pcm for videos) | null | null | null | cond-mat.soft cond-mat.stat-mech physics.chem-ph | http://creativecommons.org/licenses/by/4.0/ | Classical density functional theory (cDFT) and dynamical density functional theory (DDFT) are modern statistical mechanical theories for modeling many-body colloidal systems at the one-body density level. The theories hinge on knowing the excess free-energy accurately, which is however not feasible for most practical applications. Dijkman et al. [Phys. Rev. Lett. 134, 056103 (2025)] recently showed how a neural excess free-energy functional for cDFT can be learned from bulk simulations via pair-correlation matching. In this article, we demonstrate how this same functional can be applied to DDFT, without any retraining, to simulate non-equilibrium overdamped dynamics of inhomogeneous densities. We evaluate this on a 3D Lennard-Jones system with planar geometry under various complex external potentials and observe good agreement of the dynamical densities with those from expensive Brownian dynamic simulations, up to the limit of the adiabatic approximation. We further develop and apply an extension of DDFT based on gradient flows, to a grand-canonical system modeled after breakthrough gas adsorption studies, finding similarly good agreement. Our results demonstrate a practical route for leveraging learned free-energy functionals in DDFT, paving the way for accurate and efficient modeling of many-body non-equilibrium systems. | [
{
"version": "v1",
"created": "Wed, 14 May 2025 16:42:36 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 13:39:32 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Ram",
"Karnik",
""
],
[
"Dijkman",
"Jacobus",
""
],
[
"van Roij",
"René",
""
],
[
"van de Meent",
"Jan-Willem",
""
],
[
"Ensing",
"Bernd",
""
],
[
"Welling",
"Max",
""
],
[
"Cremers",
"Daniel",
""
]
] |
2505.09879 | Yupeng Yang | Yupeng Yang, Xinyi Dai, and Yicheng Wang | New cosmological constraints on the evolution of dark matter energy density | 11 pages, 1 figures, 2 tables. References added. PRD 111, 103534 .(submitted on 28. Feb. 2025). Comments are still welcome. In particular, if you discover that your research is relevant to our work and your findings are not currently listed in Table II of our manuscript, we would greatly appreciate it if you could contact us. We will gladly include your results in the arXiv version of Table II | null | 10.1103/8ync-vrtz | null | astro-ph.CO gr-qc | http://creativecommons.org/licenses/by/4.0/ | We constrain the evolution of dark matter energy density over time, specifically focusing on deviation from the standard model represented by the equation $\rho_{m}\propto(1+z)^{3-\varepsilon}$, where $\varepsilon$ is a constant parameter. Utilizing a diverse array of observational datasets, including baryon acoustic oscillations (BAO) data from the first release of the Dark Energy Spectroscopic Instrument (DESI), distance priors derived from cosmic microwave background (CMB) observations by the Planck satellite, Hubble rate data obtained through the cosmic chronometers (CC) method, type Ia supernova (SNIa) data from the Panthon sample, and the data from the redshift space distortion (RSD) measurements ($f\sigma_8$), we derive stringent constraints on the deviation parameter. We find that for the model under consideration, the deviation parameter is constrained to be $\varepsilon = -0.0073^{+0.0029}_{-0.0033}$, indicating a deviation of approximately $2.4\sigma$ from the scenario where dark matter and vacuum dark energy do not interact. When compared with previous studies and alternative analyses, our findings provide corroborative evidence for an interaction between dark matter and vacuum dark energy, particularly in light of the release of BAO data from DESI. | [
{
"version": "v1",
"created": "Thu, 15 May 2025 00:50:34 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 02:04:48 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Yang",
"Yupeng",
""
],
[
"Dai",
"Xinyi",
""
],
[
"Wang",
"Yicheng",
""
]
] |
2505.09933 | Ruizhe Wang | Ruizhe Wang, Ryan J. MacDonald, Neale P. Gibson, Nikole K. Lewis | Open Source High-Resolution Exoplanet Atmosphere Retrievals with POSEIDON | Documentation available at https://poseidon-retrievals.readthedocs.io/en/latest/content/retrieval_tutorials.html#high-resolution-retrievals-new | The Astronomical Journal, Volume 169 (2025), Number 6, 328 | 10.3847/1538-3881/adcac7 | null | astro-ph.EP astro-ph.IM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | High-resolution spectroscopy (R > 25,000) has opened new opportunities to characterize exoplanet atmospheres from the ground. By resolving individual lines in planetary emission and transmission spectra, one can sensitively probe the chemical inventory and temperature structure of exoplanets. However, a significant challenge to reliable and reproducible atmospheric inferences from high-resolution datasets has been the lack of open source codes for high-resolution retrievals. Here, we present a unified high-resolution retrieval framework, for both emission and transmission spectroscopy, made publicly available within the open source POSEIDON retrieval code. Our high-resolution retrieval framework is fast (typically < 12 hours), accessible (no GPUs required), and well-documented via Python notebooks. We validate our framework by reproducing previous emission retrievals of the hot Jupiter WASP-77Ab and transmission retrievals of the ultra-hot Jupiter WASP-121b. Our results are broadly consistent with those of published works when making the same data detrending assumptions, but we demonstrate that user choices can subtly propagate into retrieved chemical abundances. | [
{
"version": "v1",
"created": "Thu, 15 May 2025 03:36:39 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wang",
"Ruizhe",
""
],
[
"MacDonald",
"Ryan J.",
""
],
[
"Gibson",
"Neale P.",
""
],
[
"Lewis",
"Nikole K.",
""
]
] |
2505.10034 | Zelin Fu | Changzeng Fu, Zelin Fu, Qi Zhang, Xinhe Kuang, Jiacheng Dong, Kaifeng Su, Yikai Su, Wenbo Shi, Junfeng Yao, Yuliang Zhao, Shiqi Zhao, Jiadong Wang, Siyang Song, Chaoran Liu, Yuichiro Yoshikawa, Bj\"orn Schuller and Hiroshi Ishiguro | The First MPDD Challenge: Multimodal Personality-aware Depression Detection | This paper has been accepted as part of the MPDD Challenge in the ACMMM 2025 Grand Challenge | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Depression is a widespread mental health issue affecting diverse age groups, with notable prevalence among college students and the elderly. However, existing datasets and detection methods primarily focus on young adults, neglecting the broader age spectrum and individual differences that influence depression manifestation. Current approaches often establish a direct mapping between multimodal data and depression indicators, failing to capture the complexity and diversity of depression across individuals. This challenge includes two tracks based on age-specific subsets: Track 1 uses the MPDD-Elderly dataset for detecting depression in older adults, and Track 2 uses the MPDD-Young dataset for detecting depression in younger participants. The Multimodal Personality-aware Depression Detection (MPDD) Challenge aims to address this gap by incorporating multimodal data alongside individual difference factors. We provide a baseline model that fuses audio and video modalities with individual difference information to detect depression manifestations in diverse populations. This challenge aims to promote the development of more personalized and accurate de pression detection methods, advancing mental health research and fostering inclusive detection systems. More details are available on the official challenge website: https://hacilab.github.io/MPDDChallenge.github.io. | [
{
"version": "v1",
"created": "Thu, 15 May 2025 07:29:33 GMT"
},
{
"version": "v2",
"created": "Sun, 25 May 2025 12:12:32 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 02:12:21 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Fu",
"Changzeng",
""
],
[
"Fu",
"Zelin",
""
],
[
"Zhang",
"Qi",
""
],
[
"Kuang",
"Xinhe",
""
],
[
"Dong",
"Jiacheng",
""
],
[
"Su",
"Kaifeng",
""
],
[
"Su",
"Yikai",
""
],
[
"Shi",
"Wenbo",
""
],
[
"Yao",
"Junfeng",
""
],
[
"Zhao",
"Yuliang",
""
],
[
"Zhao",
"Shiqi",
""
],
[
"Wang",
"Jiadong",
""
],
[
"Song",
"Siyang",
""
],
[
"Liu",
"Chaoran",
""
],
[
"Yoshikawa",
"Yuichiro",
""
],
[
"Schuller",
"Björn",
""
],
[
"Ishiguro",
"Hiroshi",
""
]
] |
2505.10318 | Si-Qiang Luo | Chen-Ke Zhang, Fu-Lai Wang, Si-Qiang Luo, Xiang Liu | Electromagnetic properties of possible triple-charm molecular hexaquarks | 8 pages, 2 figures, 5 tables, published in Eur. Phys. J. C | Eur. Phys. J. C 85, 582 (2025) | 10.1140/epjc/s10052-025-14317-4 | null | hep-ph hep-ex | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this study, we investigate the radiative transitions of predicted triple-charm molecular hexaquarks, which play a significant role in understanding their overall spectroscopic properties. As experimentally measurable quantities, the radiative decay widths provide insights into the internal structures of these triple-charm molecular hexaquarks. Additionally, we calculate their corresponding magnetic moments, which, together with the radiative decay widths, offer a comprehensive picture of the electromagnetic properties of these exotic states. This information is valuable for guiding future experimental searches and advancing our understanding of these unique hadronic systems. | [
{
"version": "v1",
"created": "Thu, 15 May 2025 14:03:33 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 01:22:12 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhang",
"Chen-Ke",
""
],
[
"Wang",
"Fu-Lai",
""
],
[
"Luo",
"Si-Qiang",
""
],
[
"Liu",
"Xiang",
""
]
] |
2505.10349 | Ye Zheng | Ye Zheng, Shafizur Rahman Seeam, Yidan Hu, Rui Zhang, Yanchao Zhang | Locally Differentially Private Frequency Estimation via Joint Randomized Response | Accepted by PETS'25 (issue 3) | null | 10.56553/popets-2025-0097 | null | cs.CR | http://creativecommons.org/licenses/by/4.0/ | Local Differential Privacy (LDP) has been widely recognized as a powerful tool for providing a strong theoretical guarantee of data privacy to data contributors against an untrusted data collector. Under a typical LDP scheme, each data contributor independently randomly perturbs their data before submitting them to the data collector, which in turn infers valuable statistics about the original data from received perturbed data. Common to existing LDP mechanisms is an inherent trade-off between the level of privacy protection and data utility in the sense that strong data privacy often comes at the cost of reduced data utility. Frequency estimation based on Randomized Response (RR) is a fundamental building block of many LDP mechanisms. In this paper, we propose a novel Joint Randomized Response (JRR) mechanism based on correlated data perturbations to achieve locally differentially private frequency estimation. JRR divides data contributors into disjoint groups of two members and lets those in the same group jointly perturb their binary data to improve frequency-estimation accuracy and achieve the same level of data privacy by hiding the group membership information in contrast to the classical RR mechanism. Theoretical analysis and detailed simulation studies using both real and synthetic datasets show that JRR achieves the same level of data privacy as the classical RR mechanism while improving the frequency-estimation accuracy in the overwhelming majority of the cases by up to two orders of magnitude. | [
{
"version": "v1",
"created": "Thu, 15 May 2025 14:41:03 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zheng",
"Ye",
""
],
[
"Seeam",
"Shafizur Rahman",
""
],
[
"Hu",
"Yidan",
""
],
[
"Zhang",
"Rui",
""
],
[
"Zhang",
"Yanchao",
""
]
] |
2505.10438 | David Grasev | David Grasev | Identification and Optimal Nonlinear Control of Turbojet Engine Using Koopman Eigenfunction Model | 38 pages, 28 figures | null | null | null | cs.LG cs.SY eess.SY | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Gas turbine engines represent complex highly nonlinear dynamical systems. Deriving their physics-based models can be challenging as it requires performance characteristics, that are not always available, and one often has to make many simplifying assumptions. In this paper, the limitations of conventional experimental methods used to derive component-level and locally linear parameter-varying models are discussed and addressed by employing identification techniques based on data collected from standard engine operation under closed-loop control. The rotor dynamics were estimated using the sparse identification of nonlinear dynamics. Subsequently, the autonomous part of the dynamics was mapped into an optimally constructed Koopman eigenfunction space. The process included eigenvalue optimization using metaheuristic algorithms and temporal projection, followed by gradient-based eigenfunction identification. The resulting Koopman model was validated against an in-house reference component-level model. A globally optimal nonlinear feedback controller and a Kalman estimator were then designed in the eigenfunction space and compared to the classical and gain-scheduled proportional-integral controllers, as well as a proposed internal model control approach. The eigenmode structure allowed targeting individual modes during the optimization process, resulting in a better performance tuning. The results showed that the Koopman-based controller outperformed the other benchmark controllers in both reference tracking and disturbance rejection, under sea-level and varying flight conditions, due to its global nature. | [
{
"version": "v1",
"created": "Thu, 15 May 2025 15:55:13 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 13:02:34 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Grasev",
"David",
""
]
] |
2505.10439 | Andrew Riesen | Andrew Riesen | Interpolating Feigin-Frenkel Duality at the Critical Level to Matrices of Complex Size | Sections 4 and 9 revised for clarity. References fixed and added. Comments welcome. 54 Pages | null | null | null | math.QA math.CT math.RT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we extend Feigin-Frenkel duality at the critical level to the setting of complex rank. This is accomplished by considering the center of a vertex algebra in Deligne's interpolating categories, along with Feigin's Lie algebras of complex rank, $\mathfrak{gl}_{\lambda}$ and $\mathfrak{po}_{\lambda}$. More precisely, we define the universal affine vertex algebras associated with Lie algebras in $\underline{\mathrm{Re}}\mathrm{p}(\mathrm{GL}_{\alpha},\mathbb{F})$, $\underline{\mathrm{Re}}\mathrm{p}(\mathrm{O}_{\alpha},\mathbb{F})$ and $\underline{\mathrm{Re}}\mathrm{p}(\mathrm{Sp}_{\alpha},\mathbb{F})$, and describe their centers at the critical level explicitly by interpolating Molev's construction of Segal-Sugawara vectors. Using the formalism of Poisson vertex algebras, we identify a natural set of generators for the Drinfeld-Sokolov reduction of $\mathfrak{gl}_{\lambda}$ and $\mathfrak{po}_{\lambda}$, denoted by $\mathcal{W}(\mathfrak{gl}_{\lambda})$ and $\mathcal{W}(\mathfrak{po}_{\lambda})$, respectively. Finally, we show that the interpolated Feigin-Frenkel isomorphism maps the interpolated Segal-Sugawara vectors to these generators. | [
{
"version": "v1",
"created": "Thu, 15 May 2025 15:56:49 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 00:08:37 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Riesen",
"Andrew",
""
]
] |
2505.10489 | Ilya Deriy | Ilya Deriy, Danil Kornovan, Mihail Petrov, and Andrey Bogdanov | Optical Spintronics: Towards Optical Communication Without Energy Transfer | null | null | null | null | physics.optics | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Energy, momentum, and angular momentum are fundamental properties tied to the symmetries of space and time, with photons and other elementary particles acting as carriers of these quantities. In most optical and optoelectronic devices, energy transfer is crucial, but it often results in undesirable energy absorption. Moreover, non-reciprocal elements such as optical diodes and circulators are difficult to implement in photonics, as they typically require time-dependent perturbations, nonlinear effects, or external magnetic fields. This presents a significant barrier to the development of efficient, compact photonic technologies. We introduce the concept of optical spin current, wherein spin angular momentum is transferred by an electromagnetic field without accompanying energy transfer. This phenomenon is analogous to electron spin currents, where spin is decoupled from charge flow. Building on this principle, we propose optical spin diode and circulator -- devices that enable unidirectional propagation of spin currents while maintaining bidirectional energy flow, thus preserving reciprocity. Furthermore, we demonstrate asymmetric spin transfer between quantum dots mediated by the optical spin diode, highlighting the potential for novel optical spintronic functionalities. These findings lay the foundation for devices that leverage optical spin transfer, opening new avenues for advancements in optical spintronics. | [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:43:52 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 14:07:10 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Deriy",
"Ilya",
""
],
[
"Kornovan",
"Danil",
""
],
[
"Petrov",
"Mihail",
""
],
[
"Bogdanov",
"Andrey",
""
]
] |
2505.10754 | Manaure Francisquez | Manaure Francisquez and Petr Cagas and Akash Shukla and James Juno and Gregory W. Hammett | Conservative velocity mappings for discontinuous Galerkin kinetics | null | null | null | null | physics.plasm-ph | http://creativecommons.org/licenses/by/4.0/ | Continuum computational kinetic plasma models evolve the distribution function of a plasma species $f_s$ on a phase-space grid over time. In many problems of interest the distribution function has limited extent in velocity space, hence using a highly refined mesh everywhere would be costly and slow. Nonuniform velocity grids can reduce the computational cost by placing more degrees of freedom where $f_s$ is appreciable and fewer where it is not. In this work we introduce a first-of-its kind discontinuous Galerkin approach to nonuniform velocity-space discretization using mapped velocity coordinates. This new method is presented in the context of a gyrokinetic model used to study magnetized plasmas. We create discretizations of collisionless and collisional terms using mappings in a way that exactly conserves particles and energy. Numerical tests of such properties are presented, and we show that this new discretization can reproduce earlier gyrokinetic simulations using grids with up to 48 times fewer cells. | [
{
"version": "v1",
"created": "Thu, 15 May 2025 23:47:41 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 21:32:18 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Francisquez",
"Manaure",
""
],
[
"Cagas",
"Petr",
""
],
[
"Shukla",
"Akash",
""
],
[
"Juno",
"James",
""
],
[
"Hammett",
"Gregory W.",
""
]
] |
2505.10923 | Simeon Adebola | Simeon Adebola, Shuangyu Xie, Chung Min Kim, Justin Kerr, Bart M. van Marrewijk, Mieke van Vlaardingen, Tim van Daalen, E.N. van Loo, Jose Luis Susa Rincon, Eugen Solowjow, Rick van de Zedde, and Ken Goldberg | GrowSplat: Constructing Temporal Digital Twins of Plants with Gaussian Splats | null | null | null | null | cs.RO cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Accurate temporal reconstructions of plant growth are essential for plant phenotyping and breeding, yet remain challenging due to complex geometries, occlusions, and non-rigid deformations of plants. We present a novel framework for building temporal digital twins of plants by combining 3D Gaussian Splatting with a robust sample alignment pipeline. Our method begins by reconstructing Gaussian Splats from multi-view camera data, then leverages a two-stage registration approach: coarse alignment through feature-based matching and Fast Global Registration, followed by fine alignment with Iterative Closest Point. This pipeline yields a consistent 4D model of plant development in discrete time steps. We evaluate the approach on data from the Netherlands Plant Eco-phenotyping Center, demonstrating detailed temporal reconstructions of Sequoia and Quinoa species. Videos and Images can be seen at https://berkeleyautomation.github.io/GrowSplat/ | [
{
"version": "v1",
"created": "Fri, 16 May 2025 06:56:15 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 20:00:53 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Adebola",
"Simeon",
""
],
[
"Xie",
"Shuangyu",
""
],
[
"Kim",
"Chung Min",
""
],
[
"Kerr",
"Justin",
""
],
[
"van Marrewijk",
"Bart M.",
""
],
[
"van Vlaardingen",
"Mieke",
""
],
[
"van Daalen",
"Tim",
""
],
[
"van Loo",
"E. N.",
""
],
[
"Rincon",
"Jose Luis Susa",
""
],
[
"Solowjow",
"Eugen",
""
],
[
"van de Zedde",
"Rick",
""
],
[
"Goldberg",
"Ken",
""
]
] |
2505.11047 | Arghya Mallick | Arghya Mallick, Georgios Pantazis, Mohammad Khosravi, Peyman Mohajerin Esfahani, Sergio Grammatico | User-centric Vehicle-to-Grid Optimization with an Input Convex Neural Network-based Battery Degradation Model | null | null | null | null | eess.SY cs.LG cs.SY | http://creativecommons.org/licenses/by/4.0/ | We propose a data-driven, user-centric vehicle-to-grid (V2G) methodology based on multi-objective optimization to balance battery degradation and V2G revenue according to EV user preference. Given the lack of accurate and generalizable battery degradation models, we leverage input convex neural networks (ICNNs) to develop a data-driven degradation model trained on extensive experimental datasets. This approach enables our model to capture nonconvex dependencies on battery temperature and time while maintaining convexity with respect to the charging rate. Such a partial convexity property ensures that the second stage of our methodology remains computationally efficient. In the second stage, we integrate our data-driven degradation model into a multi-objective optimization framework to generate an optimal smart charging profile for each EV. This profile effectively balances the trade-off between financial benefits from V2G participation and battery degradation, controlled by a hyperparameter reflecting the user prioritization of battery health. Numerical simulations show the high accuracy of the ICNN model in predicting battery degradation for unseen data. Finally, we present a trade-off curve illustrating financial benefits from V2G versus losses from battery health degradation based on user preferences and showcase smart charging strategies under realistic scenarios. | [
{
"version": "v1",
"created": "Fri, 16 May 2025 09:42:00 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Mallick",
"Arghya",
""
],
[
"Pantazis",
"Georgios",
""
],
[
"Khosravi",
"Mohammad",
""
],
[
"Esfahani",
"Peyman Mohajerin",
""
],
[
"Grammatico",
"Sergio",
""
]
] |
2505.11077 | Amir Bayat | Amir Bayat, Alessandro Abate, Necmiye Ozay, Raphael M. Jungers | LLM-Enhanced Symbolic Control for Safety-Critical Applications | null | null | null | null | eess.SY cs.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Motivated by Smart Manufacturing and Industry 4.0, we introduce a framework for synthesizing Abstraction-Based Controller Design (ABCD) for reach-avoid problems from Natural Language (NL) specifications using Large Language Models (LLMs). A Code Agent interprets an NL description of the control problem and translates it into a formal language interpretable by state-of-the-art symbolic control software, while a Checker Agent verifies the correctness of the generated code and enhances safety by identifying specification mismatches. Evaluations show that the system handles linguistic variability and improves robustness over direct planning with LLMs. The proposed approach lowers the barrier to formal control synthesis by enabling intuitive, NL-based task definition while maintaining safety guarantees through automated validation. | [
{
"version": "v1",
"created": "Fri, 16 May 2025 10:08:25 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 15:22:09 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Bayat",
"Amir",
""
],
[
"Abate",
"Alessandro",
""
],
[
"Ozay",
"Necmiye",
""
],
[
"Jungers",
"Raphael M.",
""
]
] |
2505.11156 | Akira Sekiyama | N. U. Sakamoto, G. Nozue, H. Fujiwara, Y. Torii, M. Sakaguchi, T. D. Nakamura, T. Kiss, H. Sugawara, S. Tanaka, Y. Iwasaki, Y. Niwa, A. Ishikawa, T. D. Yamamoto, R. Tamura, A. Yasui, A. Sekiyama | Simultaneous probes of pseudogap and disorder by hard x-ray photoemission applied for a candidate thermoelectric Al-Pd-Ru quasicrystal | 10 pages, 8 figures | null | null | null | cond-mat.mtrl-sci cond-mat.dis-nn cond-mat.other cond-mat.str-el | http://creativecommons.org/licenses/by-nc-sa/4.0/ | The bulk electronic structure of Al-Pd-Ru quasicrystal (QC) have been investigated by hard X-ray photoemission spectroscopy (HAXPES). We have found an intrinsic pseudogap structure in which the spectral weight in the vicinity of the Fermi level (E_F) is remarkably suppressed at any photon energy. The valence-band HAXPES spectra and the asymmetry of the core-level peaks indicate a contribution of the Al sites to the electronic structure in the vicinity of E_F with less contributions from the Pd and Ru sites. The disorder effects are found in the core-level lineshapes of the Al-Pd-Ru QC, of which the widths are much broader than those of the other reference crystalline solids with less disorder. | [
{
"version": "v1",
"created": "Fri, 16 May 2025 11:58:07 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 11:26:55 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Sakamoto",
"N. U.",
""
],
[
"Nozue",
"G.",
""
],
[
"Fujiwara",
"H.",
""
],
[
"Torii",
"Y.",
""
],
[
"Sakaguchi",
"M.",
""
],
[
"Nakamura",
"T. D.",
""
],
[
"Kiss",
"T.",
""
],
[
"Sugawara",
"H.",
""
],
[
"Tanaka",
"S.",
""
],
[
"Iwasaki",
"Y.",
""
],
[
"Niwa",
"Y.",
""
],
[
"Ishikawa",
"A.",
""
],
[
"Yamamoto",
"T. D.",
""
],
[
"Tamura",
"R.",
""
],
[
"Yasui",
"A.",
""
],
[
"Sekiyama",
"A.",
""
]
] |
2505.11461 | Wesley Suttle | Wesley A Suttle, Vipul K Sharma, Brian M Sadler | Signal attenuation enables scalable decentralized multi-agent reinforcement learning over networks | 10 pages, 1 figure | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multi-agent reinforcement learning (MARL) methods typically require that agents enjoy global state observability, preventing development of decentralized algorithms and limiting scalability. Recent work has shown that, under assumptions on decaying inter-agent influence, global observability can be replaced by local neighborhood observability at each agent, enabling decentralization and scalability. Real-world applications enjoying such decay properties remain underexplored, however, despite the fact that signal power decay, or signal attenuation, due to path loss is an intrinsic feature of many problems in wireless communications and radar networks. In this paper, we show that signal attenuation enables decentralization in MARL by considering the illustrative special case of performing power allocation for target detection in a radar network. To achieve this, we propose two new constrained multi-agent Markov decision process formulations of this power allocation problem, derive local neighborhood approximations for global value function and policy gradient estimates and establish corresponding error bounds, and develop decentralized saddle point policy gradient algorithms for solving the proposed problems. Our approach, though oriented towards the specific radar network problem we consider, provides a useful model for extensions to additional problems in wireless communications and radar networks. | [
{
"version": "v1",
"created": "Fri, 16 May 2025 17:14:37 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 21:19:23 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Suttle",
"Wesley A",
""
],
[
"Sharma",
"Vipul K",
""
],
[
"Sadler",
"Brian M",
""
]
] |
2505.11616 | Noel Richardson | Noel D. Richardson, Micaela Henson, Emma P. Lieb, Corey Kehl, Ryan M. Lau, Peredur M. Williams, Michael F. Corcoran, J. R. Callingham, Andr\'e-Nicolas Chen\'e, Theodore R. Gull, Kenji Hamaguchi, Yinuo Han, Matthew J. Hankins, Grant M. Hill, Jennifer L. Hoffman, Jonathan Mackey, Anthony F. J. Moffat, Benjamin J. S. Pope, Pragati Pradhan, Christopher M. P. Russell, Andreas A. C. Sander, Nicole St-Louis, Ian R. Stevens, Peter Tuthill, Gerd Weigelt, and Ryan M.T. White | Carbon-rich dust injected into the interstellar medium by Galactic WC binaries survives for hundreds of years | accepted to ApJ | null | null | null | astro-ph.SR astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | Some carbon-rich Wolf-Rayet stars (WC stars) show an infrared excess from dust emission. Dust forms in the collision of the WC wind with a companion star's wind. As this dust is carried towards the ISM at close to the WCd wind speed and the binary continues through its orbit, a spiral structure forms around the system. The shape depends on the orbital eccentricity and period, as well as stellar parameters like mass-loss rates and terminal wind speeds. Imaging of the WCd binary WR 140 with JWST/MIRI revealed 17 concentric dust shells surrounding the binary. We present new JWST imaging of four additional WCd systems (WR 48a, WR 112, WR 125, and WR 137) that were imaged in 2024. In this analysis, we show that the dust is long-lived, detected with an age of at least 130 years, but more than 300 years in some systems. Longer duration measurements are limited by sensitivity. Regular spacing of dust features confirms the periodic nature of dust formation, consistent with a connection to binary motion. We use these images to estimate the proper motion of the dust, finding the dust to propagate out to the interstellar medium with motion comparable to the wind speed of the WC stars. In addition to these results, we observe unusual structures around WR 48a, which could represent dusty clumps shaped by photoevaporation and wind ablation like young proplyd objects. These results demonstrate that WC dust is indeed long-lived and should be accounted for in galactic dust budgets. | [
{
"version": "v1",
"created": "Fri, 16 May 2025 18:23:15 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 14:51:47 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Richardson",
"Noel D.",
""
],
[
"Henson",
"Micaela",
""
],
[
"Lieb",
"Emma P.",
""
],
[
"Kehl",
"Corey",
""
],
[
"Lau",
"Ryan M.",
""
],
[
"Williams",
"Peredur M.",
""
],
[
"Corcoran",
"Michael F.",
""
],
[
"Callingham",
"J. R.",
""
],
[
"Chené",
"André-Nicolas",
""
],
[
"Gull",
"Theodore R.",
""
],
[
"Hamaguchi",
"Kenji",
""
],
[
"Han",
"Yinuo",
""
],
[
"Hankins",
"Matthew J.",
""
],
[
"Hill",
"Grant M.",
""
],
[
"Hoffman",
"Jennifer L.",
""
],
[
"Mackey",
"Jonathan",
""
],
[
"Moffat",
"Anthony F. J.",
""
],
[
"Pope",
"Benjamin J. S.",
""
],
[
"Pradhan",
"Pragati",
""
],
[
"Russell",
"Christopher M. P.",
""
],
[
"Sander",
"Andreas A. C.",
""
],
[
"St-Louis",
"Nicole",
""
],
[
"Stevens",
"Ian R.",
""
],
[
"Tuthill",
"Peter",
""
],
[
"Weigelt",
"Gerd",
""
],
[
"White",
"Ryan M. T.",
""
]
] |
2505.11692 | Sahil Rajesh Dhayalkar | Sahil Rajesh Dhayalkar | The Geometry of ReLU Networks through the ReLU Transition Graph | 13 pages, 4 figures | null | null | null | cs.LG cs.AI cs.NE | http://creativecommons.org/licenses/by/4.0/ | We develop a novel theoretical framework for analyzing ReLU neural networks through the lens of a combinatorial object we term the ReLU Transition Graph (RTG). In this graph, each node corresponds to a linear region induced by the network's activation patterns, and edges connect regions that differ by a single neuron flip. Building on this structure, we derive a suite of new theoretical results connecting RTG geometry to expressivity, generalization, and robustness. Our contributions include tight combinatorial bounds on RTG size and diameter, a proof of RTG connectivity, and graph-theoretic interpretations of VC-dimension. We also relate entropy and average degree of the RTG to generalization error. Each theoretical result is rigorously validated via carefully controlled experiments across varied network depths, widths, and data regimes. This work provides the first unified treatment of ReLU network structure via graph theory and opens new avenues for compression, regularization, and complexity control rooted in RTG analysis. | [
{
"version": "v1",
"created": "Fri, 16 May 2025 21:00:56 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 21:46:04 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Dhayalkar",
"Sahil Rajesh",
""
]
] |
2505.11694 | Sahil Rajesh Dhayalkar | Sahil Rajesh Dhayalkar | Neural Networks as Universal Finite-State Machines: A Constructive Deterministic Finite Automaton Theory | 15 pages, 1 figure | null | null | null | cs.LG cs.AI cs.FL | http://creativecommons.org/licenses/by/4.0/ | We present a complete theoretical and empirical framework establishing feedforward neural networks as universal finite-state machines (N-FSMs). Our results prove that finite-depth ReLU and threshold networks can exactly simulate deterministic finite automata (DFAs) by unrolling state transitions into depth-wise neural layers, with formal characterizations of required depth, width, and state compression. We demonstrate that DFA transitions are linearly separable, binary threshold activations allow exponential compression, and Myhill-Nerode equivalence classes can be embedded into continuous latent spaces while preserving separability. We also formalize the expressivity boundary: fixed-depth feedforward networks cannot recognize non-regular languages requiring unbounded memory. Unlike prior heuristic or probing-based studies, we provide constructive proofs and design explicit DFA-unrolled neural architectures that empirically validate every claim. Our results bridge deep learning, automata theory, and neural-symbolic computation, offering a rigorous blueprint for how discrete symbolic processes can be realized in continuous neural systems. | [
{
"version": "v1",
"created": "Fri, 16 May 2025 21:01:34 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 21:39:10 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Dhayalkar",
"Sahil Rajesh",
""
]
] |
2505.11942 | Junhao Zheng | Junhao Zheng, Xidi Cai, Qiuke Li, Duzhen Zhang, ZhongZhi Li, Yingying Zhang, Le Song, Qianli Ma | LifelongAgentBench: Evaluating LLM Agents as Lifelong Learners | null | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | Lifelong learning is essential for intelligent agents operating in dynamic environments. Current large language model (LLM)-based agents, however, remain stateless and unable to accumulate or transfer knowledge over time. Existing benchmarks treat agents as static systems and fail to evaluate lifelong learning capabilities. We present LifelongAgentBench, the first unified benchmark designed to systematically assess the lifelong learning ability of LLM agents. It provides skill-grounded, interdependent tasks across three interactive environments, Database, Operating System, and Knowledge Graph, with automatic label verification, reproducibility, and modular extensibility. Extensive experiments reveal that conventional experience replay has limited effectiveness for LLM agents due to irrelevant information and context length constraints. We further introduce a group self-consistency mechanism that significantly improves lifelong learning performance. We hope LifelongAgentBench will advance the development of adaptive, memory-capable LLM agents. | [
{
"version": "v1",
"created": "Sat, 17 May 2025 10:09:11 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 16:16:26 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zheng",
"Junhao",
""
],
[
"Cai",
"Xidi",
""
],
[
"Li",
"Qiuke",
""
],
[
"Zhang",
"Duzhen",
""
],
[
"Li",
"ZhongZhi",
""
],
[
"Zhang",
"Yingying",
""
],
[
"Song",
"Le",
""
],
[
"Ma",
"Qianli",
""
]
] |
2505.12147 | Nikolaos Lysias Kosioris | Nikolaos-Lysias Kosioris, Sotirios Nikoletseas, Gavrilis Filios, Stefanos Panagiotou | Causal Machine Learning in IoT-based Engineering Problems: A Tool Comparison in the Case of Household Energy Consumption | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | The rapid increase in computing power and the ability to store Big Data in the infrastructure has enabled predictions in a large variety of domains by Machine Learning. However, in many cases, existing Machine Learning tools are considered insufficient or incorrect since they exploit only probabilistic dependencies rather than inference logic. Causal Machine Learning methods seem to close this gap. In this paper, two prevalent tools based on Causal Machine Learning methods are compared, as well as their mathematical underpinning background. The operation of the tools is demonstrated by examining their response to 18 queries, based on the IDEAL Household Energy Dataset, published by the University of Edinburgh. First, it was important to evaluate the causal relations assumption that allowed the use of this approach; this was based on the preexisting scientific knowledge of the domain and was implemented by use of the in-built validation tools. Results were encouraging and may easily be extended to other domains. | [
{
"version": "v1",
"created": "Sat, 17 May 2025 21:39:51 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 16:25:14 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Kosioris",
"Nikolaos-Lysias",
""
],
[
"Nikoletseas",
"Sotirios",
""
],
[
"Filios",
"Gavrilis",
""
],
[
"Panagiotou",
"Stefanos",
""
]
] |
2505.12204 | Shuo Han | Shuo Han, German Espinosa, Junda Huang, Daniel A. Dombeck, Malcolm A. MacIver, Bradly C. Stadie | Of Mice and Machines: A Comparison of Learning Between Real World Mice and RL Agents | 19 pages, ICML 2025 | null | null | null | cs.LG q-bio.NC | http://creativecommons.org/licenses/by/4.0/ | Recent advances in reinforcement learning (RL) have demonstrated impressive capabilities in complex decision-making tasks. This progress raises a natural question: how do these artificial systems compare to biological agents, which have been shaped by millions of years of evolution? To help answer this question, we undertake a comparative study of biological mice and RL agents in a predator-avoidance maze environment. Through this analysis, we identify a striking disparity: RL agents consistently demonstrate a lack of self-preservation instinct, readily risking ``death'' for marginal efficiency gains. These risk-taking strategies are in contrast to biological agents, which exhibit sophisticated risk-assessment and avoidance behaviors. Towards bridging this gap between the biological and artificial, we propose two novel mechanisms that encourage more naturalistic risk-avoidance behaviors in RL agents. Our approach leads to the emergence of naturalistic behaviors, including strategic environment assessment, cautious path planning, and predator avoidance patterns that closely mirror those observed in biological systems. | [
{
"version": "v1",
"created": "Sun, 18 May 2025 02:40:16 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 15:03:47 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Han",
"Shuo",
""
],
[
"Espinosa",
"German",
""
],
[
"Huang",
"Junda",
""
],
[
"Dombeck",
"Daniel A.",
""
],
[
"MacIver",
"Malcolm A.",
""
],
[
"Stadie",
"Bradly C.",
""
]
] |
2505.12255 | Susovan Pramanik | Susovan Pramanik (Harish-Chandra Research Institute, India) | Anisotropic Calder\'{o}n Problem for a Non-Local Second Order Elliptic Operator | 18 pages | null | null | null | math.AP | http://creativecommons.org/licenses/by/4.0/ | This paper investigates the anisotropic Calder\'{o}n problem for a non-local elliptic operator of order 2, on closed Riemannian manifolds. We demonstrate that using the Cauchy data set, we can recover the geometry of a closed Riemannian manifold up to standard gauge. | [
{
"version": "v1",
"created": "Sun, 18 May 2025 06:21:23 GMT"
},
{
"version": "v2",
"created": "Tue, 20 May 2025 06:11:28 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 01:19:19 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Pramanik",
"Susovan",
"",
"Harish-Chandra Research Institute, India"
]
] |
2505.12364 | Francesco Sala | Francesco Sala | On a Riemann-Roch formula for stacks with finite cyclotomic inertia | null | null | null | null | math.AG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | B. Toen defined a Riemann-Roch map from the rational algebraic K-theory of a tame Deligne-Mumford quotient stack to the \'etale K-theory of its inertia. He proved that this map is an isomorphism and that it is covariant with respect to proper maps. Moreover G. Vezzosi and A. Vistoli proved a decomposition theorem for the equivariant K-theory of a noetherian scheme. In this paper we give a geometric definition of the Vezzosi-Vistoli decomposition, interpreting the pieces as corresponding to the components of the cyclotomic inertia. When the map from the cyclotomic inertia to the stack is finite, we can define a Riemann-Roch map in Toen's style. We prove that this map is an isomorphism and it is covariant with respect to proper relatively tame maps; moreover in some favourable circumstances we explicitly compute its inverse map, and show that we can recover Toen's one when the stack is tame Deligne-Mumford. | [
{
"version": "v1",
"created": "Sun, 18 May 2025 10:58:20 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 17:31:53 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Sala",
"Francesco",
""
]
] |
2505.12864 | Joel Niklaus | Yu Fan, Jingwei Ni, Jakob Merane, Etienne Salimbeni, Yang Tian, Yoan Hermstr\"uwer, Yinya Huang, Mubashara Akhtar, Florian Geering, Oliver Dreyer, Daniel Brunner, Markus Leippold, Mrinmaya Sachan, Alexander Stremitzer, Christoph Engel, Elliott Ash, Joel Niklaus | LEXam: Benchmarking Legal Reasoning on 340 Law Exams | null | null | null | null | cs.CL cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Long-form legal reasoning remains a key challenge for large language models (LLMs) in spite of recent advances in test-time scaling. We introduce LEXam, a novel benchmark derived from 340 law exams spanning 116 law school courses across a range of subjects and degree levels. The dataset comprises 4,886 law exam questions in English and German, including 2,841 long-form, open-ended questions and 2,045 multiple-choice questions. Besides reference answers, the open questions are also accompanied by explicit guidance outlining the expected legal reasoning approach such as issue spotting, rule recall, or rule application. Our evaluation on both open-ended and multiple-choice questions present significant challenges for current LLMs; in particular, they notably struggle with open questions that require structured, multi-step legal reasoning. Moreover, our results underscore the effectiveness of the dataset in differentiating between models with varying capabilities. Adopting an LLM-as-a-Judge paradigm with rigorous human expert validation, we demonstrate how model-generated reasoning steps can be evaluated consistently and accurately. Our evaluation setup provides a scalable method to assess legal reasoning quality beyond simple accuracy metrics. Project page: https://lexam-benchmark.github.io/ | [
{
"version": "v1",
"created": "Mon, 19 May 2025 08:48:12 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 15:37:57 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Fan",
"Yu",
""
],
[
"Ni",
"Jingwei",
""
],
[
"Merane",
"Jakob",
""
],
[
"Salimbeni",
"Etienne",
""
],
[
"Tian",
"Yang",
""
],
[
"Hermstrüwer",
"Yoan",
""
],
[
"Huang",
"Yinya",
""
],
[
"Akhtar",
"Mubashara",
""
],
[
"Geering",
"Florian",
""
],
[
"Dreyer",
"Oliver",
""
],
[
"Brunner",
"Daniel",
""
],
[
"Leippold",
"Markus",
""
],
[
"Sachan",
"Mrinmaya",
""
],
[
"Stremitzer",
"Alexander",
""
],
[
"Engel",
"Christoph",
""
],
[
"Ash",
"Elliott",
""
],
[
"Niklaus",
"Joel",
""
]
] |
2505.13260 | Alexander Efimov | Alexander I. Efimov | Some remarks on Quillen's D\'evissage theorem | 14 pages; v2: minor corrections, typos fixed | null | null | null | math.KT math.AG math.CT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper we give a different proof of Quillen's D\'evissage theorem using Barwick's theorem of the heart. The key ingredient is a certain short exact sequence of dg categories, which is closely related with the Auslander-type construction for nilpotent extensions which was used in the papers of Kuznetsov-Lunts \cite{KL15}, Land-Tamme \cite{LT19} and the author \cite{E20}. | [
{
"version": "v1",
"created": "Mon, 19 May 2025 15:42:40 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 19:03:20 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Efimov",
"Alexander I.",
""
]
] |
2505.13455 | Von Ralph Dane Herbuela | Von Ralph Dane Marquez Herbuela and Yukie Nagai | Exploring Spatiotemporal Emotional Synchrony in Dyadic Interactions: The Role of Speech Conditions in Facial and Vocal Affective Alignment | null | null | null | null | eess.AS cs.AI | http://creativecommons.org/licenses/by/4.0/ | Understanding how humans express and synchronize emotions across multiple communication channels particularly facial expressions and speech has significant implications for emotion recognition systems and human computer interaction. Motivated by the notion that non-overlapping speech promotes clearer emotional coordination, while overlapping speech disrupts synchrony, this study examines how these conversational dynamics shape the spatial and temporal alignment of arousal and valence across facial and vocal modalities. Using dyadic interactions from the IEMOCAP dataset, we extracted continuous emotion estimates via EmoNet (facial video) and a Wav2Vec2-based model (speech audio). Segments were categorized based on speech overlap, and emotional alignment was assessed using Pearson correlation, lag adjusted analysis, and Dynamic Time Warping (DTW). Across analyses, non overlapping speech was associated with more stable and predictable emotional synchrony than overlapping speech. While zero-lag correlations were low and not statistically different, non overlapping speech showed reduced variability, especially for arousal. Lag adjusted correlations and best-lag distributions revealed clearer, more consistent temporal alignment in these segments. In contrast, overlapping speech exhibited higher variability and flatter lag profiles, though DTW indicated unexpectedly tighter alignment suggesting distinct coordination strategies. Notably, directionality patterns showed that facial expressions more often preceded speech during turn-taking, while speech led during simultaneous vocalizations. These findings underscore the importance of conversational structure in regulating emotional communication and provide new insight into the spatial and temporal dynamics of multimodal affective alignment in real world interaction. | [
{
"version": "v1",
"created": "Tue, 29 Apr 2025 14:41:55 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 10:30:48 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Herbuela",
"Von Ralph Dane Marquez",
""
],
[
"Nagai",
"Yukie",
""
]
] |
2505.13476 | Francesco D'Agostino | Francesco D'Agostino | Unified Structural Embedding of Orbifold Sigma Models | I have revisited some proofs, expanded one that was outlined, justified more some other proofs and organizing the citations a bit better | null | 10.5281/zenodo.15376431 | null | math-ph math.MP quant-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This study introduces a new unified structural framework for orbifold sigma models that incorporates twisted sectors, singularities, and smooth regions into a single algebraic object. Traditional approaches to orbifold theories often treat such sectors separately, requiring ad hoc regularizations near singularities and failing at capturing inter-sector interactions under renormalization group flow. Therefore, the scope of this study aims at resolving these limitations through the construction of a unified orbifold algebra $\mathcal{A}(X/G)$ that decomposes into idempotent-projected components corresponding to conjugacy classes of the finite group $G$ acting on the target space $X$. The formalism is shown to recover conventional sigma model results in the smooth limit where $G$ approaches the trivial group, with the internal renormalization group derivation reducing to the standard one-loop beta function proportional to the Ricci tensor. Examples demonstrate the applicability, including explicit calculations for the $\mathbb{C}/\mathbb{Z}_2$ orbifold that exhibit the decomposition into untwisted and twisted field contributions. | [
{
"version": "v1",
"created": "Fri, 09 May 2025 20:20:08 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 00:30:19 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"D'Agostino",
"Francesco",
""
]
] |
2505.13510 | Conor Rowan | Conor Rowan, Alireza Doostan | On the definition and importance of interpretability in scientific machine learning | null | null | null | null | cs.LG physics.data-an physics.hist-ph physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Though neural networks trained on large datasets have been successfully used to describe and predict many physical phenomena, there is a sense among scientists that, unlike traditional scientific models comprising simple mathematical expressions, their findings cannot be integrated into the body of scientific knowledge. Critics of machine learning's inability to produce human-understandable relationships have converged on the concept of "interpretability" as its point of departure from more traditional forms of science. As the growing interest in interpretability has shown, researchers in the physical sciences seek not just predictive models, but also to uncover the fundamental principles that govern a system of interest. However, clarity around a definition of interpretability and the precise role that it plays in science is lacking in the literature. In this work, we argue that researchers in equation discovery and symbolic regression tend to conflate the concept of sparsity with interpretability. We review key papers on interpretable machine learning from outside the scientific community and argue that, though the definitions and methods they propose can inform questions of interpretability for scientific machine learning (SciML), they are inadequate for this new purpose. Noting these deficiencies, we propose an operational definition of interpretability for the physical sciences. Our notion of interpretability emphasizes understanding of the mechanism over mathematical sparsity. Innocuous though it may seem, this emphasis on mechanism shows that sparsity is often unnecessary. It also questions the possibility of interpretable scientific discovery when prior knowledge is lacking. We believe a precise and philosophically informed definition of interpretability in SciML will help focus research efforts toward the most significant obstacles to realizing a data-driven scientific future. | [
{
"version": "v1",
"created": "Fri, 16 May 2025 20:16:14 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 18:55:13 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Rowan",
"Conor",
""
],
[
"Doostan",
"Alireza",
""
]
] |
2505.13688 | Seongsil Heo | Seongsil Heo, Calvin Murdock, Michael Proulx, Christi Miller | Gaze-Enhanced Multimodal Turn-Taking Prediction in Triadic Conversations | To appear in the Proceedings of Interspeech 2025 | null | null | null | cs.HC | http://creativecommons.org/licenses/by/4.0/ | Turn-taking prediction is crucial for seamless interactions. This study introduces a novel, lightweight framework for accurate turn-taking prediction in triadic conversations without relying on computationally intensive methods. Unlike prior approaches that either disregard gaze or treat it as a passive signal, our model integrates gaze with speaker localization, structuring it within a spatial constraint to transform it into a reliable predictive cue. Leveraging egocentric behavioral cues, our experiments demonstrate that incorporating gaze data from a single-user significantly improves prediction performance, while gaze data from multiple-users further enhances it by capturing richer conversational dynamics. This study presents a lightweight and privacy-conscious approach to support adaptive, directional sound control, enhancing speech intelligibility in noisy environments, particularly for hearing assistance in smart glasses. | [
{
"version": "v1",
"created": "Mon, 19 May 2025 19:43:58 GMT"
},
{
"version": "v2",
"created": "Thu, 22 May 2025 21:52:44 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 05:58:52 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Heo",
"Seongsil",
""
],
[
"Murdock",
"Calvin",
""
],
[
"Proulx",
"Michael",
""
],
[
"Miller",
"Christi",
""
]
] |
2505.13774 | Zidi Xiong | Zidi Xiong, Shan Chen, Zhenting Qi, Himabindu Lakkaraju | Measuring the Faithfulness of Thinking Drafts in Large Reasoning Models | null | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | Large Reasoning Models (LRMs) have significantly enhanced their capabilities in complex problem-solving by introducing a thinking draft that enables multi-path Chain-of-Thought explorations before producing final answers. Ensuring the faithfulness of these intermediate reasoning processes is crucial for reliable monitoring, interpretation, and effective control. In this paper, we propose a systematic counterfactual intervention framework to rigorously evaluate thinking draft faithfulness. Our approach focuses on two complementary dimensions: (1) Intra-Draft Faithfulness, which assesses whether individual reasoning steps causally influence subsequent steps and the final draft conclusion through counterfactual step insertions; and (2) Draft-to-Answer Faithfulness, which evaluates whether final answers are logically consistent with and dependent on the thinking draft, by perturbing the draft's concluding logic. We conduct extensive experiments across six state-of-the-art LRMs. Our findings show that current LRMs demonstrate selective faithfulness to intermediate reasoning steps and frequently fail to faithfully align with the draft conclusions. These results underscore the need for more faithful and interpretable reasoning in advanced LRMs. | [
{
"version": "v1",
"created": "Mon, 19 May 2025 23:20:24 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 19:41:14 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Xiong",
"Zidi",
""
],
[
"Chen",
"Shan",
""
],
[
"Qi",
"Zhenting",
""
],
[
"Lakkaraju",
"Himabindu",
""
]
] |
2505.13814 | Jihwan Lee | Jihwan Lee, Kevin Huang, Kleanthis Avramidis, Simon Pistrosch, Monica Gonzalez-Machorro, Yoonjeong Lee, Bj\"orn Schuller, Louis Goldstein, Shrikanth Narayanan | Articulatory Feature Prediction from Surface EMG during Speech Production | Accepted for Interspeech2025 | null | null | null | eess.AS cs.AI cs.SD | http://creativecommons.org/licenses/by/4.0/ | We present a model for predicting articulatory features from surface electromyography (EMG) signals during speech production. The proposed model integrates convolutional layers and a Transformer block, followed by separate predictors for articulatory features. Our approach achieves a high prediction correlation of approximately 0.9 for most articulatory features. Furthermore, we demonstrate that these predicted articulatory features can be decoded into intelligible speech waveforms. To our knowledge, this is the first method to decode speech waveforms from surface EMG via articulatory features, offering a novel approach to EMG-based speech synthesis. Additionally, we analyze the relationship between EMG electrode placement and articulatory feature predictability, providing knowledge-driven insights for optimizing EMG electrode configurations. The source code and decoded speech samples are publicly available. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 01:50:05 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 03:59:36 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lee",
"Jihwan",
""
],
[
"Huang",
"Kevin",
""
],
[
"Avramidis",
"Kleanthis",
""
],
[
"Pistrosch",
"Simon",
""
],
[
"Gonzalez-Machorro",
"Monica",
""
],
[
"Lee",
"Yoonjeong",
""
],
[
"Schuller",
"Björn",
""
],
[
"Goldstein",
"Louis",
""
],
[
"Narayanan",
"Shrikanth",
""
]
] |
2505.13929 | Huateng Zhu | Jerome Droniou, Kim-Ngan Le, Huateng Zhu | Numerical analysis of the regularised total variation flow | null | null | null | null | math.NA cs.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We perform the numerical analysis of the regularised total variation flow using the gradient discretisation method (GDM). GDM is a unified convergence analysis framework that covers conforming and non-conforming numerical methods, for instance, conforming and non-conforming finite element, two-point flux approximation, etc.. In this paper, a fully discretised implicit scheme of the model is proposed, the existence and uniqueness of the solution to the scheme is proved, the stability and consistency of the scheme are analysed, and error estimates are established. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 04:49:33 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 11:18:38 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Droniou",
"Jerome",
""
],
[
"Le",
"Kim-Ngan",
""
],
[
"Zhu",
"Huateng",
""
]
] |
2505.14020 | Hao Dong | Hao Dong, Ziyue Qiao, Zhiyuan Ning, Qi Hao, Yi Du, Pengyang Wang, Yuanchun Zhou | Disentangled Multi-span Evolutionary Network against Temporal Knowledge Graph Reasoning | Accepted to ACL 2025 Findings | null | null | null | cs.AI cs.IR cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Temporal Knowledge Graphs (TKGs), as an extension of static Knowledge Graphs (KGs), incorporate the temporal feature to express the transience of knowledge by describing when facts occur. TKG extrapolation aims to infer possible future facts based on known history, which has garnered significant attention in recent years. Some existing methods treat TKG as a sequence of independent subgraphs to model temporal evolution patterns, demonstrating impressive reasoning performance. However, they still have limitations: 1) In modeling subgraph semantic evolution, they usually neglect the internal structural interactions between subgraphs, which are actually crucial for encoding TKGs. 2) They overlook the potential smooth features that do not lead to semantic changes, which should be distinguished from the semantic evolution process. Therefore, we propose a novel Disentangled Multi-span Evolutionary Network (DiMNet) for TKG reasoning. Specifically, we design a multi-span evolution strategy that captures local neighbor features while perceiving historical neighbor semantic information, thus enabling internal interactions between subgraphs during the evolution process. To maximize the capture of semantic change patterns, we design a disentangle component that adaptively separates nodes' active and stable features, used to dynamically control the influence of historical semantics on future evolution. Extensive experiments conducted on four real-world TKG datasets show that DiMNet demonstrates substantial performance in TKG reasoning, and outperforms the state-of-the-art up to 22.7% in MRR. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 07:22:03 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 07:45:13 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Dong",
"Hao",
""
],
[
"Qiao",
"Ziyue",
""
],
[
"Ning",
"Zhiyuan",
""
],
[
"Hao",
"Qi",
""
],
[
"Du",
"Yi",
""
],
[
"Wang",
"Pengyang",
""
],
[
"Zhou",
"Yuanchun",
""
]
] |
2505.14030 | Zhiqian Lan | Zhiqian Lan, Yuxuan Jiang, Ruiqi Wang, Xuanbing Xie, Rongkui Zhang, Yicheng Zhu, Peihang Li, Tianshuo Yang, Tianxing Chen, Haoyu Gao, Xiaokang Yang, Xuelong Li, Hongyuan Zhang, Yao Mu, Ping Luo | AutoBio: A Simulation and Benchmark for Robotic Automation in Digital Biology Laboratory | null | null | null | null | cs.RO | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Vision-language-action (VLA) models have shown promise as generalist robotic policies by jointly leveraging visual, linguistic, and proprioceptive modalities to generate action trajectories. While recent benchmarks have advanced VLA research in domestic tasks, professional science-oriented domains remain underexplored. We introduce AutoBio, a simulation framework and benchmark designed to evaluate robotic automation in biology laboratory environments--an application domain that combines structured protocols with demanding precision and multimodal interaction. AutoBio extends existing simulation capabilities through a pipeline for digitizing real-world laboratory instruments, specialized physics plugins for mechanisms ubiquitous in laboratory workflows, and a rendering stack that support dynamic instrument interfaces and transparent materials through physically based rendering. Our benchmark comprises biologically grounded tasks spanning three difficulty levels, enabling standardized evaluation of language-guided robotic manipulation in experimental protocols. We provide infrastructure for demonstration generation and seamless integration with VLA models. Baseline evaluations with two SOTA VLA models reveal significant gaps in precision manipulation, visual reasoning, and instruction following in scientific workflows. By releasing AutoBio, we aim to catalyze research on generalist robotic systems for complex, high-precision, and multimodal professional environments. The simulator and benchmark are publicly available to facilitate reproducible research. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 07:29:26 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 16:17:21 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 01:50:49 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lan",
"Zhiqian",
""
],
[
"Jiang",
"Yuxuan",
""
],
[
"Wang",
"Ruiqi",
""
],
[
"Xie",
"Xuanbing",
""
],
[
"Zhang",
"Rongkui",
""
],
[
"Zhu",
"Yicheng",
""
],
[
"Li",
"Peihang",
""
],
[
"Yang",
"Tianshuo",
""
],
[
"Chen",
"Tianxing",
""
],
[
"Gao",
"Haoyu",
""
],
[
"Yang",
"Xiaokang",
""
],
[
"Li",
"Xuelong",
""
],
[
"Zhang",
"Hongyuan",
""
],
[
"Mu",
"Yao",
""
],
[
"Luo",
"Ping",
""
]
] |
2505.14107 | Yakun Zhu | Yakun Zhu, Zhongzhen Huang, Linjie Mu, Yutong Huang, Wei Nie, Jiaji Liu, Shaoting Zhang, Pengfei Liu, Xiaofan Zhang | DiagnosisArena: Benchmarking Diagnostic Reasoning for Large Language Models | null | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The emergence of groundbreaking large language models capable of performing complex reasoning tasks holds significant promise for addressing various scientific challenges, including those arising in complex clinical scenarios. To enable their safe and effective deployment in real-world healthcare settings, it is urgently necessary to benchmark the diagnostic capabilities of current models systematically. Given the limitations of existing medical benchmarks in evaluating advanced diagnostic reasoning, we present DiagnosisArena, a comprehensive and challenging benchmark designed to rigorously assess professional-level diagnostic competence. DiagnosisArena consists of 1,113 pairs of segmented patient cases and corresponding diagnoses, spanning 28 medical specialties, deriving from clinical case reports published in 10 top-tier medical journals. The benchmark is developed through a meticulous construction pipeline, involving multiple rounds of screening and review by both AI systems and human experts, with thorough checks conducted to prevent data leakage. Our study reveals that even the most advanced reasoning models, o3, o1, and DeepSeek-R1, achieve only 51.12%, 31.09%, and 17.79% accuracy, respectively. This finding highlights a significant generalization bottleneck in current large language models when faced with clinical diagnostic reasoning challenges. Through DiagnosisArena, we aim to drive further advancements in AI's diagnostic reasoning capabilities, enabling more effective solutions for real-world clinical diagnostic challenges. We provide the benchmark and evaluation tools for further research and development https://github.com/SPIRAL-MED/DiagnosisArena. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 09:14:53 GMT"
},
{
"version": "v2",
"created": "Sat, 24 May 2025 12:44:04 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 03:43:01 GMT"
},
{
"version": "v4",
"created": "Thu, 29 May 2025 08:24:00 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhu",
"Yakun",
""
],
[
"Huang",
"Zhongzhen",
""
],
[
"Mu",
"Linjie",
""
],
[
"Huang",
"Yutong",
""
],
[
"Nie",
"Wei",
""
],
[
"Liu",
"Jiaji",
""
],
[
"Zhang",
"Shaoting",
""
],
[
"Liu",
"Pengfei",
""
],
[
"Zhang",
"Xiaofan",
""
]
] |
2505.14279 | Jennifer D'Souza | Jennifer D'Souza, Hamed Babaei Giglou and Quentin M\"unch | YESciEval: Robust LLM-as-a-Judge for Scientific Question Answering | 9 pages, 4 figures, Accepted as a Long Paper at the 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025) | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by-sa/4.0/ | Large Language Models (LLMs) drive scientific question-answering on modern search engines, yet their evaluation robustness remains underexplored. We introduce YESciEval, an open-source framework that combines fine-grained rubric-based assessment with reinforcement learning to mitigate optimism bias in LLM evaluators. We release multidisciplinary scienceQ&A datasets, including adversarial variants, with evaluation scores from multiple LLMs. Independent of proprietary models and human feedback, our approach enables scalable, cost-free evaluation. By advancing reliable LLM-as-a-judge models, this work supports AI alignment and fosters robust, transparent evaluation essential for scientific inquiry. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 12:30:46 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 16:45:00 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"D'Souza",
"Jennifer",
""
],
[
"Giglou",
"Hamed Babaei",
""
],
[
"Münch",
"Quentin",
""
]
] |
2505.14359 | Xu Chen | Ruoxin Chen, Junwei Xi, Zhiyuan Yan, Ke-Yue Zhang, Shuang Wu, Jingyi Xie, Xu Chen, Lei Xu, Isabel Guan, Taiping Yao, Shouhong Ding | Dual Data Alignment Makes AI-Generated Image Detector Easier Generalizable | 12 Pages, 9 figures | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Existing detectors are often trained on biased datasets, leading to the possibility of overfitting on non-causal image attributes that are spuriously correlated with real/synthetic labels. While these biased features enhance performance on the training data, they result in substantial performance degradation when applied to unbiased datasets. One common solution is to perform dataset alignment through generative reconstruction, matching the semantic content between real and synthetic images. However, we revisit this approach and show that pixel-level alignment alone is insufficient. The reconstructed images still suffer from frequency-level misalignment, which can perpetuate spurious correlations. To illustrate, we observe that reconstruction models tend to restore the high-frequency details lost in real images (possibly due to JPEG compression), inadvertently creating a frequency-level misalignment, where synthetic images appear to have richer high-frequency content than real ones. This misalignment leads to models associating high-frequency features with synthetic labels, further reinforcing biased cues. To resolve this, we propose Dual Data Alignment (DDA), which aligns both the pixel and frequency domains. Moreover, we introduce two new test sets: DDA-COCO, containing DDA-aligned synthetic images for testing detector performance on the most aligned dataset, and EvalGEN, featuring the latest generative models for assessing detectors under new generative architectures such as visual auto-regressive generators. Finally, our extensive evaluations demonstrate that a detector trained exclusively on DDA-aligned MSCOCO could improve across 8 diverse benchmarks by a non-trivial margin, showing a +7.2% on in-the-wild benchmarks, highlighting the improved generalizability of unbiased detectors. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 13:42:38 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 12:59:23 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 13:36:34 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Chen",
"Ruoxin",
""
],
[
"Xi",
"Junwei",
""
],
[
"Yan",
"Zhiyuan",
""
],
[
"Zhang",
"Ke-Yue",
""
],
[
"Wu",
"Shuang",
""
],
[
"Xie",
"Jingyi",
""
],
[
"Chen",
"Xu",
""
],
[
"Xu",
"Lei",
""
],
[
"Guan",
"Isabel",
""
],
[
"Yao",
"Taiping",
""
],
[
"Ding",
"Shouhong",
""
]
] |
2505.14456 | Oksana Dobush Dr | O.A. Dobush, M.P. Kozlovskii, I.V. Pylyuk, Yu.O. Plevachuk | Influence of microscopic parameters on phase behavior of a cell model with Curie-Weiss interaction | 11 pages, 2 figures | null | null | null | cond-mat.stat-mech | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate how varying two microscopic parameters - cell volume and the ratio between repulsion and attraction intensities - affect the phase behavior of a cell model with a Curie-Weiss-type interaction. The analysis is based on an exact solution previously derived for this model in the grand canonical ensemble. At sufficiently low temperatures, the cell model exhibits multiple first-order phase transitions. By varying the cell volume and the repulsion-to-attraction ratio, we represent a quantitative comparison of the chemical potential and pressure isotherms, along with the pressure-temperature and temperature-density phase diagrams. Our results demonstrate that altering these microscopic parameters induces quantitative changes in the phase diagrams of the cell model. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 14:53:34 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 14:15:50 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Dobush",
"O. A.",
""
],
[
"Kozlovskii",
"M. P.",
""
],
[
"Pylyuk",
"I. V.",
""
],
[
"Plevachuk",
"Yu. O.",
""
]
] |
2505.14528 | Hongrong Yin | Hongrong Yin, Jinhong Huang, Yao Li, Yunwei Dong, Tao Zhang | BugRepro: Enhancing Android Bug Reproduction with Domain-Specific Knowledge Integration | null | null | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Mobile application development is a fast-paced process where maintaining high-quality user experiences is crucial. Bug reproduction, a key aspect of maintaining app quality, often faces significant challenges. Specifically, when descriptions in bug reports are ambiguous or difficult to comprehend, current approaches fail to extract accurate information. Moreover, modern applications exhibit inherent complexity with multiple pages and diverse functionalities, making it challenging for existing methods to map the relevant information in bug reports to the corresponding UI elements that need to be manipulated. To address these challenges, we propose BugRepro, a novel technique that integrates domain-specific knowledge to enhance the accuracy and efficiency of bug reproduction. BugRepro adopts a Retrieval-Augmented Generation (RAG) approach. It retrieves similar bug reports along with their corresponding steps to reproduce (S2R) entities from an example-rich RAG document. In addition, BugRepro explores the graphical user interface (GUI) of the app and extracts transition graphs from the user interface to incorporate app-specific knowledge to guide large language models (LLMs) in their exploration process. Our experiments demonstrate that BugRepro significantly outperforms two state-of-the-art methods (ReCDroid and AdbGPT). For S2R entity extraction accuracy, it achieves a 7.57 to 28.89 percentage point increase over prior methods. For the bug reproduction success rate, the improvement reaches 74.55% and 152.63%. In reproduction efficiency, the gains are 0.72% and 76.68%. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 15:48:34 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 13:03:01 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Yin",
"Hongrong",
""
],
[
"Huang",
"Jinhong",
""
],
[
"Li",
"Yao",
""
],
[
"Dong",
"Yunwei",
""
],
[
"Zhang",
"Tao",
""
]
] |
2505.14669 | Andrei Panferov | Roberto L. Castro, Andrei Panferov, Soroush Tabesh, Oliver Sieberling, Jiale Chen, Mahdi Nikdan, Saleh Ashkboos, Dan Alistarh | Quartet: Native FP4 Training Can Be Optimal for Large Language Models | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Training large language models (LLMs) models directly in low-precision offers a way to address computational costs by improving both throughput and energy efficiency. For those purposes, NVIDIA's recent Blackwell architecture facilitates very low-precision operations using FP4 variants. Yet, current algorithms for training LLMs in FP4 precision face significant accuracy degradation and often rely on mixed-precision fallbacks. In this paper, we investigate hardware-supported FP4 training and introduce a new approach for accurate, end-to-end FP4 training with all the major computations (i.e., linear layers) in low precision. Through extensive evaluations on Llama-type models, we reveal a new low-precision scaling law that quantifies performance trade-offs across bit-widths and training setups. Guided by this investigation, we design an "optimal" technique in terms of accuracy-vs-computation, called Quartet. We implement Quartet using optimized CUDA kernels tailored for Blackwell, demonstrating that fully FP4-based training is a competitive alternative to FP16 half-precision and to FP8 training. Our code is available at https://github.com/IST-DASLab/Quartet. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 17:55:50 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 16:32:48 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Castro",
"Roberto L.",
""
],
[
"Panferov",
"Andrei",
""
],
[
"Tabesh",
"Soroush",
""
],
[
"Sieberling",
"Oliver",
""
],
[
"Chen",
"Jiale",
""
],
[
"Nikdan",
"Mahdi",
""
],
[
"Ashkboos",
"Saleh",
""
],
[
"Alistarh",
"Dan",
""
]
] |
2505.14762 | Jiaxin Zhang | Nikolai Makarov, Jiaxin Zhang | Multiple radial SLE($\kappa$) and quantum Calogero-Sutherland system | 48 pages, 9 pages. First part of a split from arXiv:2410.21544v1. Second part is updated as arXiv:2410.21544v2 | null | null | null | math.PR math-ph math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this first of two articles, we study the multiple radial $\mathrm{SLE}(\kappa)$ systems with parameter $\kappa > 0$ -- a family of random multi-curve systems in a simply-connected domain $\Omega$, with marked boundary points $z_1, \ldots, z_n \in \partial \Omega$ and a marked interior point $q$. As a consequence of the domain Markov property and conformal invariance, we show that such systems are characterized by equivalence classes of partition functions, which are not necessarily conformally covariant. Nevertheless, within each equivalence class, one can always choose a conformally covariant representative. When the marked interior point $q$ is set at the origin, we demonstrate that the partition function satisfies a system of second-order PDEs, known as the null vector equations, with a null vector constant $h$ and a rotation equation involving a constant $\omega$. Motivated by the Coulomb gas formalism in conformal field theory, we construct four families of solutions to the null vector equations, which are naturally classified according to topological link patterns. For $\kappa > 0$, the partition functions of multiple radial SLE($\kappa$) systems correspond to eigenstates of the quantum Calogero-Sutherland (CS) Hamiltonian beyond the states built upon the fermionic states. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 17:17:01 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 02:05:16 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Makarov",
"Nikolai",
""
],
[
"Zhang",
"Jiaxin",
""
]
] |
2505.14788 | Valeria Coenda | Valeria Coenda, Gustavo Baume, Tali Palma, Carlos Feinstein | Global properties, fractality, and mass segregation in single, paired, and grouped open clusters | Accepted for publication in A&A | null | null | null | astro-ph.GA astro-ph.SR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate the structural and dynamical properties of Open Clusters (OCs) classified as single, in pairs, or in groups. By analysing their mass, size, age, fractality, and mass segregation, we aim to identify systematic differences among these categories and evaluate the role of the Galactic environment in their evolution. Our sample comprises 420 single OCs, 415 in pairs, and 317 in groups. To characterise their structure, we apply the Q-parameter, which distinguishes fractal from radial distributions. We also compute the local density ratio to quantify mass segregation and explore its dependence on environment. Grouped OCs tend to be the youngest, followed by those in pairs, while single OCs generally exhibit older ages. Although sizes are comparable, OCs in pairs and groups tend to be less concentrated. Structurally, grouped OCs show the highest fractality, which decreases with age as clusters evolve towards more radial configurations. Mass segregation is detected in ~80% of OCs, with a slightly higher incidence in single clusters. Some older single OCs show inverse segregation, with massive stars at larger radii. Spatially, single OCs are more dispersed, whereas paired and grouped ones are concentrated in spiral arms and star-forming regions. OC evolution appears to be shaped by both internal dynamics and environmental influences. Single OCs tend to exhibit signs of more advanced dynamical evolution, whereas those in pairs and groups may retain features reflecting their formation environment. Substructures and high fractality in younger clusters suggest that early interactions play a key role in their long-term development. More massive OCs evolve towards radial configurations, while less massive ones may retain fractal properties for longer. These findings highlight the interplay between intrinsic properties and external conditions in shaping OC evolution. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 18:00:08 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 13:49:37 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Coenda",
"Valeria",
""
],
[
"Baume",
"Gustavo",
""
],
[
"Palma",
"Tali",
""
],
[
"Feinstein",
"Carlos",
""
]
] |
2505.15057 | Frederic Wang | Frederic Wang, Jonathan I. Tamir | Non-rigid Motion Correction for MRI Reconstruction via Coarse-To-Fine Diffusion Models | ICIP 2025 | null | null | null | eess.IV cs.CV | http://creativecommons.org/licenses/by/4.0/ | Magnetic Resonance Imaging (MRI) is highly susceptible to motion artifacts due to the extended acquisition times required for k-space sampling. These artifacts can compromise diagnostic utility, particularly for dynamic imaging. We propose a novel alternating minimization framework that leverages a bespoke diffusion model to jointly reconstruct and correct non-rigid motion-corrupted k-space data. The diffusion model uses a coarse-to-fine denoising strategy to capture large overall motion and reconstruct the lower frequencies of the image first, providing a better inductive bias for motion estimation than that of standard diffusion models. We demonstrate the performance of our approach on both real-world cine cardiac MRI datasets and complex simulated rigid and non-rigid deformations, even when each motion state is undersampled by a factor of 64x. Additionally, our method is agnostic to sampling patterns, anatomical variations, and MRI scanning protocols, as long as some low frequency components are sampled during each motion state. | [
{
"version": "v1",
"created": "Wed, 21 May 2025 03:27:21 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 06:38:03 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wang",
"Frederic",
""
],
[
"Tamir",
"Jonathan I.",
""
]
] |
2505.15552 | Ziwen Song | Yujuan Liu, Ziwen Song, Tingting Lin, Biao Tang and Aoxing Hao | Enhanced Robustness of Atom Interferometer Using Super-Gaussian Pulses | null | null | null | null | physics.atom-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Laser frequency drifts and atomic thermal motion can lead to errors in pulse duration and detuning in cold atom interferometry, thereby reducing measurement stability and fringe contrast. To address this issue, we investigate the use of super-Gaussian pulses, which are characterized by smooth temporal profiles and centralized energy distribution, in the beam-splitting and reflection stages of an atom interferometer. Through numerical simulations, we compare the performance of rectangular, Gaussian, and 2nd- to 10th-order super-Gaussian pulses subject to deviations in pulse duration and detuning. Our results show that super-Gaussian pulses significantly enhance interference fringe contrast and robustness, with 4th-order pulses achieving up to a 90% improvement in contrast over rectangular pulses under realistic conditions. These findings demonstrate the potential of super-Gaussian pulse shaping to enhance the sensitivity and robustness of atom interferometric measurements. | [
{
"version": "v1",
"created": "Wed, 21 May 2025 14:14:14 GMT"
},
{
"version": "v2",
"created": "Thu, 22 May 2025 12:03:57 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 10:42:23 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Liu",
"Yujuan",
""
],
[
"Song",
"Ziwen",
""
],
[
"Lin",
"Tingting",
""
],
[
"Tang",
"Biao",
""
],
[
"Hao",
"Aoxing",
""
]
] |
2505.15726 | Anton Nazarov | Anton Nazarov and Anton Selemenchuk | Fluctuations of Young diagrams for symplectic groups and semiclassical orthogonal polynomials | 29 pages, 1 figure, 1 table. v2 fixes incorrect statement about tensor power and normalization of polynomials, adds table of polynomials | null | null | null | math.PR math-ph math.MP math.RT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Consider an $n\times k$ matrix of i.i.d. Bernoulli random numbers with $p=1/2$. Dual RSK algorithm gives a bijection of this matrix to a pair of Young tableaux of conjugate shape, which is manifestation of skew Howe $GL_{n}\times GL_{k}$-duality. Thus the probability measure on zero-ones matrix leads to the probability measure on Young diagrams proportional to the ratio of the dimension of $GL_{n}\times GL_{k}$-representation and the dimension of the exterior algebra $\bigwedge\left(\mathbb{C}^{n}\otimes\mathbb{C}^{k}\right)$.
Similarly, by applying Proctor's algorithm based on Berele's modification of the Schensted insertion, we get skew Howe duality for the pairs of groups $Sp_{2n}\times Sp_{2k}$. In the limit when $n,k\to\infty$ $GL$-case is relatively easily studied by use of free-fermionic representation for the correlation kernel. But for the symplectic groups there is no convenient free-fermionic representation. We use Christoffel transformation to obtain the semiclassical orthogonal polynomials for $Sp_{2n}\times Sp_{2k}$ from Krawtchouk polynomials that describe $GL_{2n}\times GL_{2k}$ case. We derive an integral representation for semiclassical polynomials. The study of the asymptotic of this integral representation gives us the description of the limit shapes and fluctuations of the random Young diagrams for symplectic groups. | [
{
"version": "v1",
"created": "Wed, 21 May 2025 16:33:12 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 17:34:12 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Nazarov",
"Anton",
""
],
[
"Selemenchuk",
"Anton",
""
]
] |
2505.15750 | Jack Kelly | Jack Kelly, Devarshi Mukherjee | Localising invariants in derived bornological geometry | v1 - 96 pages, comments and feedback very welcome!; v2- Some corrections; references added, new material on passable morphisms of stacks (including detail about certain categories of sheaves being rigid) | null | null | null | math.KT math.AG math.AT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study several categories of analytic stacks relative to the category of bornological modules over a Banach ring. When the underlying Banach ring is a nonarchimedean valued field, this category contains derived rigid analytic spaces as a full subcategory. When the underlying field is the complex numbers, it contains the category of derived complex analytic spaces. In the second part of the paper, we consider localising invariants of rigid categories associated to bornological algebras. The main results in this part include Nisnevich descent for derived analytic spaces and a version of the Grothendieck-Riemann-Roch Theorem for derived dagger analytic spaces over an arbitrary Banach ring. | [
{
"version": "v1",
"created": "Wed, 21 May 2025 16:57:37 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 17:58:02 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Kelly",
"Jack",
""
],
[
"Mukherjee",
"Devarshi",
""
]
] |
2505.15799 | Jake Hofman | David M. Rothschild and Markus Mobius and Jake M. Hofman and Eleanor W. Dillon and Daniel G. Goldstein and Nicole Immorlica and Sonia Jaffe and Brendan Lucier and Aleksandrs Slivkins and Matthew Vogel | The Agentic Economy | null | null | null | null | cs.CY | http://creativecommons.org/licenses/by/4.0/ | Generative AI has transformed human-computer interaction by enabling natural language interfaces and the emergence of autonomous agents capable of acting on users' behalf. While early applications have improved individual productivity, these gains have largely been confined to predefined tasks within existing workflows. We argue that the more profound economic impact lies in reducing communication frictions between consumers and businesses. This shift could reorganize markets, redistribute power, and catalyze the creation of new products and services. We explore the implications of an agentic economy, where assistant agents act on behalf of consumers and service agents represent businesses, interacting programmatically to facilitate transactions. A key distinction we draw is between unscripted interactions -- enabled by technical advances in natural language and protocol design -- and unrestricted interactions, which depend on market structures and governance. We examine the current limitations of siloed and end-to-end agents, and explore future scenarios shaped by technical standards and market dynamics. These include the potential tension between agentic walled gardens and an open web of agents, implications for advertising and discovery, the evolution of micro-transactions, and the unbundling and rebundling of digital goods. Ultimately, we argue that the architecture of agentic communication will determine the extent to which generative AI democratizes access to economic opportunity. | [
{
"version": "v1",
"created": "Wed, 21 May 2025 17:51:36 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 17:44:50 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Rothschild",
"David M.",
""
],
[
"Mobius",
"Markus",
""
],
[
"Hofman",
"Jake M.",
""
],
[
"Dillon",
"Eleanor W.",
""
],
[
"Goldstein",
"Daniel G.",
""
],
[
"Immorlica",
"Nicole",
""
],
[
"Jaffe",
"Sonia",
""
],
[
"Lucier",
"Brendan",
""
],
[
"Slivkins",
"Aleksandrs",
""
],
[
"Vogel",
"Matthew",
""
]
] |
2505.15886 | Sami Kaya | Raphael Bousso and Sami Kaya | Fundamental Complement of a Gravitating Region | null | null | null | null | hep-th gr-qc quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Any gravitating region $a$ in any spacetime gives rise to a generalized entanglement wedge, the hologram $e(a)$. Holograms exhibit properties expected of fundamental operator algebras, such as strong subadditivity, nesting, and no-cloning. But the entanglement wedge EW of an AdS boundary region $B$ with commutant $\bar B$ satisfies an additional condition, complementarity: EW$(B)$ is the spacelike complement of EW$(\bar B)$ in the bulk.
Here we identify an analogue of the boundary commutant $\bar B$ in general spacetimes: given a gravitating region $a$, its \emph{fundamental complement} $\tilde{a}$ is the smallest wedge that contains all infinite world lines contained in the spacelike complement $a'$ of $a$. We refine the definition of $e(a)$ by requiring that it be spacelike to $\tilde a$. We prove that $e(a)$ is the spacelike complement of $e(\tilde a)$ when the latter is computed in $a'$.
We exhibit many examples of $\tilde{a}$ and of $e(a)$ in de Sitter, flat, and cosmological spacetimes. We find that a Big Bang cosmology (spatially closed or not) is trivially reconstructible: the whole universe is the entanglement wedge of any wedge inside it. But de Sitter space is not trivially reconstructible, despite being closed. We recover the AdS/CFT prescription by proving that EW$(B)=e($causal wedge of $B$). | [
{
"version": "v1",
"created": "Wed, 21 May 2025 18:00:00 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 23:50:42 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Bousso",
"Raphael",
""
],
[
"Kaya",
"Sami",
""
]
] |
2505.15914 | YuanKuei Wu | Yuan-Kuei Wu, Juan Azcarreta, Kashyap Patel, Buye Xu, Jung-Suk Lee, Sanha Lee, Ashutosh Pandey | A Novel Deep Learning Framework for Efficient Multichannel Acoustic Feedback Control | Accepted by Interspeech 2025 | null | null | null | cs.SD eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This study presents a deep-learning framework for controlling multichannel acoustic feedback in audio devices. Traditional digital signal processing methods struggle with convergence when dealing with highly correlated noise such as feedback. We introduce a Convolutional Recurrent Network that efficiently combines spatial and temporal processing, significantly enhancing speech enhancement capabilities with lower computational demands. Our approach utilizes three training methods: In-a-Loop Training, Teacher Forcing, and a Hybrid strategy with a Multichannel Wiener Filter, optimizing performance in complex acoustic environments. This scalable framework offers a robust solution for real-world applications, making significant advances in Acoustic Feedback Control technology. | [
{
"version": "v1",
"created": "Wed, 21 May 2025 18:07:48 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 13:45:47 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wu",
"Yuan-Kuei",
""
],
[
"Azcarreta",
"Juan",
""
],
[
"Patel",
"Kashyap",
""
],
[
"Xu",
"Buye",
""
],
[
"Lee",
"Jung-Suk",
""
],
[
"Lee",
"Sanha",
""
],
[
"Pandey",
"Ashutosh",
""
]
] |
2505.15929 | Hui Shen | Hui Shen, Taiqiang Wu, Qi Han, Yunta Hsieh, Jizhou Wang, Yuyue Zhang, Yuxin Cheng, Zijian Hao, Yuansheng Ni, Xin Wang, Zhongwei Wan, Kai Zhang, Wendong Xu, Jing Xiong, Ping Luo, Wenhu Chen, Chaofan Tao, Zhuoqing Mao, Ngai Wong | PhyX: Does Your Model Have the "Wits" for Physical Reasoning? | null | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Existing benchmarks fail to capture a crucial aspect of intelligence: physical reasoning, the integrated ability to combine domain knowledge, symbolic reasoning, and understanding of real-world constraints. To address this gap, we introduce PhyX: the first large-scale benchmark designed to assess models capacity for physics-grounded reasoning in visual scenarios. PhyX includes 3K meticulously curated multimodal questions spanning 6 reasoning types across 25 sub-domains and 6 core physics domains: thermodynamics, electromagnetism, mechanics, modern physics, optics, and wave\&acoustics. In our comprehensive evaluation, even state-of-the-art models struggle significantly with physical reasoning. GPT-4o, Claude3.7-Sonnet, and GPT-o4-mini achieve only 32.5%, 42.2%, and 45.8% accuracy respectively-performance gaps exceeding 29% compared to human experts. Our analysis exposes critical limitations in current models: over-reliance on memorized disciplinary knowledge, excessive dependence on mathematical formulations, and surface-level visual pattern matching rather than genuine physical understanding. We provide in-depth analysis through fine-grained statistics, detailed case studies, and multiple evaluation paradigms to thoroughly examine physical reasoning capabilities. To ensure reproducibility, we implement a compatible evaluation protocol based on widely-used toolkits such as VLMEvalKit, enabling one-click evaluation. More details are available on our project page: https://phyx-bench.github.io/. | [
{
"version": "v1",
"created": "Wed, 21 May 2025 18:33:50 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 17:59:14 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Shen",
"Hui",
""
],
[
"Wu",
"Taiqiang",
""
],
[
"Han",
"Qi",
""
],
[
"Hsieh",
"Yunta",
""
],
[
"Wang",
"Jizhou",
""
],
[
"Zhang",
"Yuyue",
""
],
[
"Cheng",
"Yuxin",
""
],
[
"Hao",
"Zijian",
""
],
[
"Ni",
"Yuansheng",
""
],
[
"Wang",
"Xin",
""
],
[
"Wan",
"Zhongwei",
""
],
[
"Zhang",
"Kai",
""
],
[
"Xu",
"Wendong",
""
],
[
"Xiong",
"Jing",
""
],
[
"Luo",
"Ping",
""
],
[
"Chen",
"Wenhu",
""
],
[
"Tao",
"Chaofan",
""
],
[
"Mao",
"Zhuoqing",
""
],
[
"Wong",
"Ngai",
""
]
] |
2505.16091 | Jinpei Guo | Jinpei Guo, Yifei Ji, Zheng Chen, Kai Liu, Min Liu, Wang Rao, Wenbo Li, Yong Guo, and Yulun Zhang | OSCAR: One-Step Diffusion Codec for Image Compression Across Multiple Bit-rates | null | null | null | null | eess.IV cs.CV | http://creativecommons.org/licenses/by/4.0/ | Pretrained latent diffusion models have shown strong potential for lossy image compression, owing to their powerful generative priors. Most existing diffusion-based methods reconstruct images by iteratively denoising from random noise, guided by compressed latent representations. While these approaches have achieved high reconstruction quality, their multi-step sampling process incurs substantial computational overhead. Moreover, they typically require training separate models for different compression bit-rates, leading to significant training and storage costs. To address these challenges, we propose a one-step diffusion codec across multiple bit-rates. termed OSCAR. Specifically, our method views compressed latents as noisy variants of the original latents, where the level of distortion depends on the bit-rate. This perspective allows them to be modeled as intermediate states along a diffusion trajectory. By establishing a mapping from the compression bit-rate to a pseudo diffusion timestep, we condition a single generative model to support reconstructions at multiple bit-rates. Meanwhile, we argue that the compressed latents retain rich structural information, thereby making one-step denoising feasible. Thus, OSCAR replaces iterative sampling with a single denoising pass, significantly improving inference efficiency. Extensive experiments demonstrate that OSCAR achieves superior performance in both quantitative and visual quality metrics. The code and models will be released at https://github.com/jp-guo/OSCAR. | [
{
"version": "v1",
"created": "Thu, 22 May 2025 00:14:12 GMT"
},
{
"version": "v2",
"created": "Sat, 24 May 2025 12:28:19 GMT"
},
{
"version": "v3",
"created": "Wed, 28 May 2025 20:58:12 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Guo",
"Jinpei",
""
],
[
"Ji",
"Yifei",
""
],
[
"Chen",
"Zheng",
""
],
[
"Liu",
"Kai",
""
],
[
"Liu",
"Min",
""
],
[
"Rao",
"Wang",
""
],
[
"Li",
"Wenbo",
""
],
[
"Guo",
"Yong",
""
],
[
"Zhang",
"Yulun",
""
]
] |
2505.16093 | Jiaxin Zhang | Jiaxin Zhang | Multiple chordal SLE($\kappa$) and quantum Calogero-Moser system | 21 pages, 1 figure. arXiv admin note: substantial text overlap with arXiv:2410.21544, arXiv:2505.14762 | null | null | null | math.PR math-ph math.CV math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study multiple chordal SLE$(\kappa)$ systems in a simply connected domain $\Omega$, where $z_1, \ldots, z_n \in \partial \Omega$ are boundary starting points, and $q \in \partial \Omega$ is an additional marked boundary point.
Based on the framework introduced in \cite{Dub07}, we show that the partition functions in the upper half-plane $\mathbb{H}$ satisfy both the null vector equations and a dilatation equation with scaling constant $d$.
We show that the presence of the marked boundary point $q$ introduces a natural equivalence relation among partition functions, which are not necessarily conformally covariant. Nevertheless, within each equivalence class, a conformally covariant representative can always be chosen.
Using techniques from the Coulomb gas formalism in conformal field theory, we construct two distinct families of solutions, each indexed by a topological link pattern of type $(n,m)$, where $2m \leq n$.
In the special case where $\Omega = \mathbb{H}$ and $q = \infty$, we demonstrate that these partition functions correspond to eigenstates of the quantum Calogero-Moser system, extending the known correspondence beyond the standard $(2n,n)$ setting. | [
{
"version": "v1",
"created": "Thu, 22 May 2025 00:24:32 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 02:18:06 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhang",
"Jiaxin",
""
]
] |
2505.16223 | Sangyong Lee | Sangyong Lee, Subo Hwang, Dohoon Kim | MADCluster: Model-agnostic Anomaly Detection with Self-supervised Clustering Network | 24 pages, 9 figures | null | null | null | cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we propose MADCluster, a novel model-agnostic anomaly detection framework utilizing self-supervised clustering. MADCluster is applicable to various deep learning architectures and addresses the 'hypersphere collapse' problem inherent in existing deep learning-based anomaly detection methods. The core idea is to cluster normal pattern data into a 'single cluster' while simultaneously learning the cluster center and mapping data close to this center. Also, to improve expressiveness and enable effective single clustering, we propose a new 'One-directed Adaptive loss'. The optimization of this loss is mathematically proven. MADCluster consists of three main components: Base Embedder capturing high-dimensional temporal dynamics, Cluster Distance Mapping, and Sequence-wise Clustering for continuous center updates. Its model-agnostic characteristics are achieved by applying various architectures to the Base Embedder. Experiments on four time series benchmark datasets demonstrate that applying MADCluster improves the overall performance of comparative models. In conclusion, the compatibility of MADCluster shows potential for enhancing model performance across various architectures. | [
{
"version": "v1",
"created": "Thu, 22 May 2025 04:50:44 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 07:21:39 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lee",
"Sangyong",
""
],
[
"Hwang",
"Subo",
""
],
[
"Kim",
"Dohoon",
""
]
] |
2505.16235 | Kosei Nakagawa | Kosei Nakagawa and Yoshiko Kanada-En'yo | $\alpha+{}^2n+{}^2n$ 3-body cluster structures and dineutron breaking in ${}^8 \mathrm{He}$ | 11 pages, 5 figures, 6 tables | null | null | null | nucl-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The $0_2^+$ state of ${}^8\mathrm{He}$ has been discovered by a recent experiment, which suggested a developed cluster structure of spatially correlated neutron pairs, called "dineutrons" (${}^2n$). We aim to investigate the structure of ${}^8\mathrm{He}(0^+_1)$ and ${}^8\mathrm{He}(0^+_2)$ to clarify the monopole excitation mode in ${}^8\mathrm{He}$ system while focusing on the $\alpha+{}^2n+{}^2n$ cluster structures and the breaking of ${}^2n$ clusters. We apply a microscopic cluster model with the generator coordinate method for the $\alpha+{}^2n+{}^2n$ cluster and ${}^6\mathrm{He}+{}^2n$ cluster dynamics. The $p_{3/2}$-closure component, which is induced by the spin-orbit force, is also incorporated. The present calculation reasonably reproduces the experimental data of the properties of ${}^8\mathrm{He}$, such as $2n$ and $4n$ separation energies, energy spectra, and radii. The spatially developed cluster structure of the $0_2^+$ state is obtained. The $0_1^+$ and $0_2^+$ states have dominant components of $\alpha+{}^2n+{}^2n$ 3-body cluster, but contain significant ${}^2n$ breaking components which contribute to the energy gain and size shrinking of the ${}^8\mathrm{He}$ system. The monopole excitation in ${}^8\mathrm{He}$ is regarded as a radial excitation, which is similar to that in ${}^{12}\mathrm{C}$. The $\alpha+{}^2n+{}^2n$ cluster structures play a dominant role in ${}^8\mathrm{He}(0_{1,2}^+)$, and the mixing of the dineutron breaking contributes to significant structural effects. | [
{
"version": "v1",
"created": "Thu, 22 May 2025 05:10:00 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 03:29:35 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Nakagawa",
"Kosei",
""
],
[
"Kanada-En'yo",
"Yoshiko",
""
]
] |
2505.16403 | Huazi Pan | Huazi Pan, Yanjun Zhang, Leo Yu Zhang, Scott Adams, Abbas Kouzani, Suiyang Khoo | Performance Guaranteed Poisoning Attacks in Federated Learning: A Sliding Mode Approach | This paper is to appear in IJCAI 2025, code available at: https://github.com/Halsey777/FedSA | null | null | null | cs.LG cs.SY eess.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Manipulation of local training data and local updates, i.e., the poisoning attack, is the main threat arising from the collaborative nature of the federated learning (FL) paradigm. Most existing poisoning attacks aim to manipulate local data/models in a way that causes denial-of-service (DoS) issues. In this paper, we introduce a novel attack method, named Federated Learning Sliding Attack (FedSA) scheme, aiming at precisely introducing the extent of poisoning in a subtle controlled manner. It operates with a predefined objective, such as reducing global model's prediction accuracy by 10%. FedSA integrates robust nonlinear control-Sliding Mode Control (SMC) theory with model poisoning attacks. It can manipulate the updates from malicious clients to drive the global model towards a compromised state, achieving this at a controlled and inconspicuous rate. Additionally, leveraging the robust control properties of FedSA allows precise control over the convergence bounds, enabling the attacker to set the global accuracy of the poisoned model to any desired level. Experimental results demonstrate that FedSA can accurately achieve a predefined global accuracy with fewer malicious clients while maintaining a high level of stealth and adjustable learning rates. | [
{
"version": "v1",
"created": "Thu, 22 May 2025 08:54:17 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 00:20:42 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Pan",
"Huazi",
""
],
[
"Zhang",
"Yanjun",
""
],
[
"Zhang",
"Leo Yu",
""
],
[
"Adams",
"Scott",
""
],
[
"Kouzani",
"Abbas",
""
],
[
"Khoo",
"Suiyang",
""
]
] |
2505.16499 | Roberto Morabito | Roberto Morabito, SiYoung Jang | Smaller, Smarter, Closer: The Edge of Collaborative Generative AI | This paper has been accepted for publication in IEEE Internet Computing. Upon publication, the copyright will be transferred to IEEE | null | null | null | cs.DC cs.AI cs.NI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The rapid adoption of generative AI (GenAI), particularly Large Language Models (LLMs), has exposed critical limitations of cloud-centric deployments, including latency, cost, and privacy concerns. Meanwhile, Small Language Models (SLMs) are emerging as viable alternatives for resource-constrained edge environments, though they often lack the capabilities of their larger counterparts. This article explores the potential of collaborative inference systems that leverage both edge and cloud resources to address these challenges. By presenting distinct cooperation strategies alongside practical design principles and experimental insights, we offer actionable guidance for deploying GenAI across the computing continuum. | [
{
"version": "v1",
"created": "Thu, 22 May 2025 10:34:48 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 09:04:02 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Morabito",
"Roberto",
""
],
[
"Jang",
"SiYoung",
""
]
] |
2505.16508 | Roberto Morabito | SiYoung Jang, Roberto Morabito | Edge-First Language Model Inference: Models, Metrics, and Tradeoffs | This paper has been accepted for publication and presentation at the 45th IEEE International Conference on Distributed Computing Systems (IEEE ICDCS 2025). The copyright will be transferred to IEEE upon publication in the conference proceedings | null | null | null | cs.DC cs.AI cs.NI cs.PF | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The widespread adoption of Language Models (LMs) across industries is driving interest in deploying these services across the computing continuum, from the cloud to the network edge. This shift aims to reduce costs, lower latency, and improve reliability and privacy. Small Language Models (SLMs), enabled by advances in model compression, are central to this shift, offering a path to on-device inference on resource-constrained edge platforms. This work examines the interplay between edge and cloud deployments, starting from detailed benchmarking of SLM capabilities on single edge devices, and extending to distributed edge clusters. We identify scenarios where edge inference offers comparable performance with lower costs, and others where cloud fallback becomes essential due to limits in scalability or model capacity. Rather than proposing a one-size-fits-all solution, we present platform-level comparisons and design insights for building efficient, adaptive LM inference systems across heterogeneous environments. | [
{
"version": "v1",
"created": "Thu, 22 May 2025 10:43:00 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 08:56:27 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Jang",
"SiYoung",
""
],
[
"Morabito",
"Roberto",
""
]
] |
2505.16512 | Liu Jiaxin | Jiaxin Liu, Jia Wang, Saihui Hou, Min Ren, Huijia Wu, Zhaofeng He | Beyond Face Swapping: A Diffusion-Based Digital Human Benchmark for Multimodal Deepfake Detection | null | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In recent years, the rapid development of deepfake technology has given rise to an emerging and serious threat to public security: diffusion-based digital human generation. Unlike traditional face manipulation methods, such models can generate highly realistic videos with consistency through multimodal control signals. Their flexibility and covertness pose severe challenges to existing detection strategies. To bridge this gap, we introduce DigiFakeAV, the new large-scale multimodal digital human forgery dataset based on diffusion models. Employing five of the latest digital human generation methods and the voice cloning methods, we systematically produce a dataset comprising 60,000 videos (8.4 million frames), covering multiple nationalities, skin tones, genders, and real-world scenarios, significantly enhancing data diversity and realism. User studies demonstrate that participants misclassify forged videos as real in 68% of tests, and existing detection models exhibit a large drop in performance on DigiFakeAV, highlighting the challenge of the dataset. To address this problem, we propose DigiShield, an effective detection baseline based on spatiotemporal and cross-modal fusion. By jointly modeling the 3D spatiotemporal features of videos and the semantic-acoustic features of audio, DigiShield achieves state-of-the-art (SOTA) performance on the DigiFakeAV and shows strong generalization on other datasets. | [
{
"version": "v1",
"created": "Thu, 22 May 2025 10:46:37 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 04:00:33 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Liu",
"Jiaxin",
""
],
[
"Wang",
"Jia",
""
],
[
"Hou",
"Saihui",
""
],
[
"Ren",
"Min",
""
],
[
"Wu",
"Huijia",
""
],
[
"He",
"Zhaofeng",
""
]
] |
2505.16883 | Etera R. Livine | Alexandre Arbey, Etera R. Livine, Clara Montagnon | Test-Field vs Physical Quasi-Normal Modes in Scalar-Tensor Theories | 34 pages, several plots, v2: corrected affiliations, added references | null | null | null | gr-qc | http://creativecommons.org/licenses/by/4.0/ | In the context of the general effort to model black hole dynamics, and in particular their return-to-equilibrium through quasi-normal modes, it is crucial to understand how much test-field perturbations deviate from physical perturbations in modified gravity scenarios. On the one hand, physical perturbations follow the modified Einstein equations of the considered extension of general relativity. The complexity of those equations can quickly escalate with extra fields and non-linear couplings. On the other hand, test-field perturbations, with negligible back-reaction on the space-time geometry, describe the propagation of both matter fields and spin $s=2$ gravitational waves on the black hole geometry. They are not subject to the intricacies of the modified Einstein equations, and only probe the background spacetime metric. If their physics were to not deviate significantly from physical perturbations, they would be especially useful to investigate predictions from quantum gravity scenarios which lack explicit detailed Einstein equations.
Here we focus on a specific modified gravity solution -- BCL black holes in scalar-tensor theories -- for which physical perturbations and related QNM frequencies have already been studied and computed numerically. We compute the test-field QNM frequencies and compare the two QNM spectra. This provides a concrete example of the significant differences arising between test-fields and physical perturbations, and flags unphysical deviations related to the test-field framework. | [
{
"version": "v1",
"created": "Thu, 22 May 2025 16:36:53 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 16:46:47 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Arbey",
"Alexandre",
""
],
[
"Livine",
"Etera R.",
""
],
[
"Montagnon",
"Clara",
""
]
] |
2505.16968 | Ahmed Heakl | Ahmed Heakl, Sarim Hashmi, Gustavo Bertolo Stahl, Seung Hun Eddie Han, Salman Khan, Abdulrahman Mahmoud | CASS: Nvidia to AMD Transpilation with Data, Models, and Benchmark | 20 pages, 11 figures, 5 tables | null | null | null | cs.AR cs.AI cs.CL cs.LG cs.PL | http://creativecommons.org/licenses/by/4.0/ | We introduce CASS, the first large-scale dataset and model suite for cross-architecture GPU code transpilation, targeting both source-level (CUDA <--> HIP) and assembly-level (Nvidia SASS <--> AMD RDNA3) translation. The dataset comprises 70k verified code pairs across host and device, addressing a critical gap in low-level GPU code portability. Leveraging this resource, we train the CASS family of domain-specific language models, achieving 95% source translation accuracy and 37.5% assembly translation accuracy, substantially outperforming commercial baselines such as GPT-4o, Claude, and Hipify. Our generated code matches native performance in over 85% of test cases, preserving runtime and memory behavior. To support rigorous evaluation, we introduce CASS-Bench, a curated benchmark spanning 16 GPU domains with ground-truth execution. All data, models, and evaluation tools are released as open source to foster progress in GPU compiler tooling, binary compatibility, and LLM-guided hardware translation. | [
{
"version": "v1",
"created": "Thu, 22 May 2025 17:48:53 GMT"
},
{
"version": "v2",
"created": "Sun, 25 May 2025 16:51:39 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 05:44:32 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Heakl",
"Ahmed",
""
],
[
"Hashmi",
"Sarim",
""
],
[
"Stahl",
"Gustavo Bertolo",
""
],
[
"Han",
"Seung Hun Eddie",
""
],
[
"Khan",
"Salman",
""
],
[
"Mahmoud",
"Abdulrahman",
""
]
] |
2505.16983 | Junlong Tong | Junlong Tong, Jinlan Fu, Zixuan Lin, Yingqi Fan, Anhao Zhao, Hui Su, Xiaoyu Shen | LLM as Effective Streaming Processor: Bridging Streaming-Batch Mismatches with Group Position Encoding | ACL 2025 Findings | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Large Language Models (LLMs) are primarily designed for batch processing. Existing methods for adapting LLMs to streaming rely either on expensive re-encoding or specialized architectures with limited scalability. This work identifies three key mismatches in adapting batch-oriented LLMs to streaming: (1) input-attention, (2) output-attention, and (3) position-ID mismatches. While it is commonly assumed that the latter two mismatches require frequent re-encoding, our analysis reveals that only the input-attention mismatch significantly impacts performance, indicating re-encoding outputs is largely unnecessary. To better understand this discrepancy with the common assumption, we provide the first comprehensive analysis of the impact of position encoding on LLMs in streaming, showing that preserving relative positions within source and target contexts is more critical than maintaining absolute order. Motivated by the above analysis, we introduce a group position encoding paradigm built on batch architectures to enhance consistency between streaming and batch modes. Extensive experiments on cross-lingual and cross-modal tasks demonstrate that our method outperforms existing approaches. Our method requires no architectural modifications, exhibits strong generalization in both streaming and batch modes. The code is available at repository https://github.com/EIT-NLP/StreamingLLM. | [
{
"version": "v1",
"created": "Thu, 22 May 2025 17:53:28 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 13:22:18 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Tong",
"Junlong",
""
],
[
"Fu",
"Jinlan",
""
],
[
"Lin",
"Zixuan",
""
],
[
"Fan",
"Yingqi",
""
],
[
"Zhao",
"Anhao",
""
],
[
"Su",
"Hui",
""
],
[
"Shen",
"Xiaoyu",
""
]
] |
2505.17139 | Wanghan Xu | Wanghan Xu, Xiangyu Zhao, Yuhao Zhou, Xiaoyu Yue, Ben Fei, Fenghua Ling, Wenlong Zhang, Lei Bai | EarthSE: A Benchmark Evaluating Earth Scientific Exploration Capability for Large Language Models | null | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Advancements in Large Language Models (LLMs) drive interest in scientific applications, necessitating specialized benchmarks such as Earth science. Existing benchmarks either present a general science focus devoid of Earth science specificity or cover isolated subdomains, lacking holistic evaluation. Furthermore, current benchmarks typically neglect the assessment of LLMs' capabilities in open-ended scientific exploration. In this paper, we present a comprehensive and professional benchmark for the Earth sciences, designed to evaluate the capabilities of LLMs in scientific exploration within this domain, spanning from fundamental to advanced levels. Leveraging a corpus of 100,000 research papers, we first construct two Question Answering (QA) datasets: Earth-Iron, which offers extensive question coverage for broad assessment, and Earth-Silver, which features a higher level of difficulty to evaluate professional depth. These datasets encompass five Earth spheres, 114 disciplines, and 11 task categories, assessing foundational knowledge crucial for scientific exploration. Most notably, we introduce Earth-Gold with new metrics, a dataset comprising open-ended multi-turn dialogues specifically designed to evaluate the advanced capabilities of LLMs in scientific exploration, including methodology induction, limitation analysis, and concept proposal. Extensive experiments reveal limitations in 11 leading LLMs across different domains and tasks, highlighting considerable room for improvement in their scientific exploration capabilities. The benchmark is available on https://huggingface.co/ai-earth . | [
{
"version": "v1",
"created": "Thu, 22 May 2025 06:46:08 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 06:12:51 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Xu",
"Wanghan",
""
],
[
"Zhao",
"Xiangyu",
""
],
[
"Zhou",
"Yuhao",
""
],
[
"Yue",
"Xiaoyu",
""
],
[
"Fei",
"Ben",
""
],
[
"Ling",
"Fenghua",
""
],
[
"Zhang",
"Wenlong",
""
],
[
"Bai",
"Lei",
""
]
] |
2505.17159 | Teepanis Chachiyo | Teepanis Chachiyo | Simple and accurate complete elliptic integrals for the full range of modulus | Comments are welcome. Add AGM implementation for the exct integrals | null | null | null | physics.gen-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The complete elliptic integral of the first and second kind, K(k) and E(k), appear in a multitude of physics and engineering applications. Because there is no known closed-form, the exact values have to be computed numerically. Here, approximations for the integrals are proposed based on their asymptotic behaviors. An inverse of K is also presented. As a result, the proposed K(k) and E(k) reproduce the exact analytical forms both in the zero and asymptotic limits, while in the mid-range of modulus maintain average error of 0.06% and 0.01% respectively. The key finding is the ability to compute the integrals with exceptional accuracy on both limits of elliptical conditions. An accuracy of 1 in 1,000 should be sufficient for practical or prototyping engineering and architecture designs. The simplicity should facilitate discussions of advanced physics topics in introductory physics classes, and enable broader collaborations among researchers from other fields of expertise. For example, the phase space of energy-conserving nonlinear pendulum using only elementary functions is discussed. The proposed inverse of K is shown to be Never Failing Newton Initialization and is an important step for the computation of the exact inverse. An algorithm based on Arithmetic-Geometric Mean for computing exact integrals and their derivatives are also presented, which should be useful in a platform that special functions are not accessible such as web-based and firmware developments. | [
{
"version": "v1",
"created": "Thu, 22 May 2025 14:42:51 GMT"
},
{
"version": "v2",
"created": "Mon, 26 May 2025 17:49:11 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 17:53:27 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Chachiyo",
"Teepanis",
""
]
] |
2505.17365 | Rohan Ghuge | Rohan Ghuge and Vidya Muthukumar and Sahil Singla | Improved and Oracle-Efficient Online $\ell_1$-Multicalibration | Accepted to ICML 2025 | null | null | null | cs.LG cs.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study \emph{online multicalibration}, a framework for ensuring calibrated predictions across multiple groups in adversarial settings, across $T$ rounds. Although online calibration is typically studied in the $\ell_1$ norm, prior approaches to online multicalibration have taken the indirect approach of obtaining rates in other norms (such as $\ell_2$ and $\ell_{\infty}$) and then transferred these guarantees to $\ell_1$ at additional loss. In contrast, we propose a direct method that achieves improved and oracle-efficient rates of $\widetilde{\mathcal{O}}(T^{-1/3})$ and $\widetilde{\mathcal{O}}(T^{-1/4})$ respectively, for online $\ell_1$-multicalibration. Our key insight is a novel reduction of online \(\ell_1\)-multicalibration to an online learning problem with product-based rewards, which we refer to as \emph{online linear-product optimization} ($\mathtt{OLPO}$).
To obtain the improved rate of $\widetilde{\mathcal{O}}(T^{-1/3})$, we introduce a linearization of $\mathtt{OLPO}$ and design a no-regret algorithm for this linearized problem. Although this method guarantees the desired sublinear rate (nearly matching the best rate for online calibration), it is computationally expensive when the group family \(\mathcal{H}\) is large or infinite, since it enumerates all possible groups. To address scalability, we propose a second approach to $\mathtt{OLPO}$ that makes only a polynomial number of calls to an offline optimization (\emph{multicalibration evaluation}) oracle, resulting in \emph{oracle-efficient} online \(\ell_1\)-multicalibration with a rate of $\widetilde{\mathcal{O}}(T^{-1/4})$. Our framework also extends to certain infinite families of groups (e.g., all linear functions on the context space) by exploiting a $1$-Lipschitz property of the \(\ell_1\)-multicalibration error with respect to \(\mathcal{H}\). | [
{
"version": "v1",
"created": "Fri, 23 May 2025 00:37:49 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 02:21:58 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Ghuge",
"Rohan",
""
],
[
"Muthukumar",
"Vidya",
""
],
[
"Singla",
"Sahil",
""
]
] |
2505.17366 | Yichi Zhang | Yichi Zhang, Zhihao Duan, Yuning Huang, Fengqing Zhu | Low-Rank Adaptation of Pre-trained Vision Backbones for Energy-Efficient Image Coding for Machine | 2025 IEEE International Conference on Image Processing (ICIP2025). Fix typo | null | null | null | eess.IV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Image Coding for Machines (ICM) focuses on optimizing image compression for AI-driven analysis rather than human perception. Existing ICM frameworks often rely on separate codecs for specific tasks, leading to significant storage requirements, training overhead, and computational complexity. To address these challenges, we propose an energy-efficient framework that leverages pre-trained vision backbones to extract robust and versatile latent representations suitable for multiple tasks. We introduce a task-specific low-rank adaptation mechanism, which refines the pre-trained features to be both compressible and tailored to downstream applications. This design minimizes trainable parameters and reduces energy costs for multi-task scenarios. By jointly optimizing task performance and entropy minimization, our method enables efficient adaptation to diverse tasks and datasets without full fine-tuning, achieving high coding efficiency. Extensive experiments demonstrate that our framework significantly outperforms traditional codecs and pre-processors, offering an energy-efficient and effective solution for ICM applications. The code and the supplementary materials will be available at: https://gitlab.com/viper-purdue/efficient-compression. | [
{
"version": "v1",
"created": "Fri, 23 May 2025 00:40:10 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 00:23:44 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhang",
"Yichi",
""
],
[
"Duan",
"Zhihao",
""
],
[
"Huang",
"Yuning",
""
],
[
"Zhu",
"Fengqing",
""
]
] |
2505.17418 | Rudrajit Choudhuri | Rudrajit Choudhuri, Bianca Trinkenreich, Rahul Pandita, Eirini Kalliamvakou, Igor Steinmacher, Marco Gerosa, Christopher Sanchez, Anita Sarma | What Needs Attention? Prioritizing Drivers of Developers' Trust and Adoption of Generative AI | Journal extension of ICSE 2025 paper (arXiv:2409.04099) | null | null | null | cs.HC cs.SE | http://creativecommons.org/licenses/by/4.0/ | Generative AI (genAI) tools are advertised as productivity aids. Yet, issues related to miscalibrated trust and usage friction continue to hinder their adoption. Additionally, AI can be exclusionary, failing to support diverse users adequately, further exacerbating these concerns. One such aspect of diversity is cognitive diversity -- variations in users' cognitive styles -- that leads to divergence in interaction styles. When an individual's cognitive styles are unsupported, it creates additional barriers to technology adoption. Thus, to design tools that developers trust, we must first understand what factors affect their trust and intentions to use these tools in practice?
We developed a theoretical model of factors influencing trust and adoption intentions towards genAI through a large-scale survey with developers (N=238) at GitHub and Microsoft. Using Partial Least Squares-Structural Equation Modeling (PLS-SEM), we found that genAI's system/output quality, functional value, and goal maintenance significantly influence developers' trust, which along with their cognitive styles, affects their intentions to use these tools in work. An Importance-Performance Matrix Analysis (IPMA) identified factors that, despite their strong influence, underperform, revealing specific genAI aspects that need design prioritization. We bolster these findings by qualitatively analyzing developers' perceived challenges and risks of genAI usage to uncover why these gaps persist in development contexts. For genAI to indeed be a true productivity aid rather than a disguised productivity sink, it must align with developers' goals, maintain contextual transparency, reduce cognitive burden, and provide equitable interaction support. We provide practical suggestions to guide future genAI tool design for effective, trustworthy, and inclusive human-genAI interactions. | [
{
"version": "v1",
"created": "Fri, 23 May 2025 03:05:56 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 19:36:52 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Choudhuri",
"Rudrajit",
""
],
[
"Trinkenreich",
"Bianca",
""
],
[
"Pandita",
"Rahul",
""
],
[
"Kalliamvakou",
"Eirini",
""
],
[
"Steinmacher",
"Igor",
""
],
[
"Gerosa",
"Marco",
""
],
[
"Sanchez",
"Christopher",
""
],
[
"Sarma",
"Anita",
""
]
] |
2505.17473 | Jiangning Zhu | Jiangning Zhu, Yuxing Zhou, Zheng Wang, Juntao Yao, Yima Gu, Yuhui Yuan, Shixia Liu | OrionBench: A Benchmark for Chart and Human-Recognizable Object Detection in Infographics | null | null | null | null | cs.CV cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Given the central role of charts in scientific, business, and communication contexts, enhancing the chart understanding capabilities of vision-language models (VLMs) has become increasingly critical. A key limitation of existing VLMs lies in their inaccurate visual grounding of infographic elements, including charts and human-recognizable objects (HROs) such as icons and images. However, chart understanding often requires identifying relevant elements and reasoning over them. To address this limitation, we introduce OrionBench, a benchmark designed to support the development of accurate object detection models for charts and HROs in infographics. It contains 26,250 real and 78,750 synthetic infographics, with over 6.9 million bounding box annotations. These annotations are created by combining the model-in-the-loop and programmatic methods. We demonstrate the usefulness of OrionBench through three applications: 1) constructing a Thinking-with-Boxes scheme to boost the chart understanding performance of VLMs, 2) comparing existing object detection models, and 3) applying the developed detection model to document layout and UI element detection. | [
{
"version": "v1",
"created": "Fri, 23 May 2025 04:56:07 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 16:03:54 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 03:56:55 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhu",
"Jiangning",
""
],
[
"Zhou",
"Yuxing",
""
],
[
"Wang",
"Zheng",
""
],
[
"Yao",
"Juntao",
""
],
[
"Gu",
"Yima",
""
],
[
"Yuan",
"Yuhui",
""
],
[
"Liu",
"Shixia",
""
]
] |
2505.17652 | Deyang Kong | Deyang Kong, Qi Guo, Xiangyu Xi, Wei Wang, Jingang Wang, Xunliang Cai, Shikun Zhang, Wei Ye | Rethinking the Sampling Criteria in Reinforcement Learning for LLM Reasoning: A Competence-Difficulty Alignment Perspective | null | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Reinforcement learning exhibits potential in enhancing the reasoning abilities of large language models, yet it is hard to scale for the low sample efficiency during the rollout phase. Existing methods attempt to improve efficiency by scheduling problems based on problem difficulties. However, these approaches suffer from unstable and biased estimations of problem difficulty and fail to capture the alignment between model competence and problem difficulty in RL training, leading to suboptimal results. To tackle these limitations, this paper introduces $\textbf{C}$ompetence-$\textbf{D}$ifficulty $\textbf{A}$lignment $\textbf{S}$ampling ($\textbf{CDAS}$), which enables accurate and stable estimation of problem difficulties by aggregating historical performance discrepancies of problems. Then the model competence is quantified to adaptively select problems whose difficulty is in alignment with the model's current competence using a fixed-point system. Experimental results across a range of challenging mathematical benchmarks show that CDAS achieves great improvements in both accuracy and efficiency. CDAS attains the highest average accuracy against baselines and exhibits significant speed advantages compared to Dynamic Sampling, a competitive strategy in DAPO, which is 2.33 times slower than CDAS. | [
{
"version": "v1",
"created": "Fri, 23 May 2025 09:15:26 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 11:48:49 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Kong",
"Deyang",
""
],
[
"Guo",
"Qi",
""
],
[
"Xi",
"Xiangyu",
""
],
[
"Wang",
"Wei",
""
],
[
"Wang",
"Jingang",
""
],
[
"Cai",
"Xunliang",
""
],
[
"Zhang",
"Shikun",
""
],
[
"Ye",
"Wei",
""
]
] |
2505.17656 | Hexiang Tan | Hexiang Tan, Fei Sun, Sha Liu, Du Su, Qi Cao, Xin Chen, Jingang Wang, Xunliang Cai, Yuanzhuo Wang, Huawei Shen, Xueqi Cheng | Too Consistent to Detect: A Study of Self-Consistent Errors in LLMs | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | As large language models (LLMs) often generate plausible but incorrect content, error detection has become increasingly critical to ensure truthfulness. However, existing detection methods often overlook a critical problem we term as self-consistent error, where LLMs repeatly generate the same incorrect response across multiple stochastic samples. This work formally defines self-consistent errors and evaluates mainstream detection methods on them. Our investigation reveals two key findings: (1) Unlike inconsistent errors, whose frequency diminishes significantly as LLM scale increases, the frequency of self-consistent errors remains stable or even increases. (2) All four types of detection methshods significantly struggle to detect self-consistent errors. These findings reveal critical limitations in current detection methods and underscore the need for improved methods. Motivated by the observation that self-consistent errors often differ across LLMs, we propose a simple but effective cross-model probe method that fuses hidden state evidence from an external verifier LLM. Our method significantly enhances performance on self-consistent errors across three LLM families. | [
{
"version": "v1",
"created": "Fri, 23 May 2025 09:18:56 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 06:51:44 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Tan",
"Hexiang",
""
],
[
"Sun",
"Fei",
""
],
[
"Liu",
"Sha",
""
],
[
"Su",
"Du",
""
],
[
"Cao",
"Qi",
""
],
[
"Chen",
"Xin",
""
],
[
"Wang",
"Jingang",
""
],
[
"Cai",
"Xunliang",
""
],
[
"Wang",
"Yuanzhuo",
""
],
[
"Shen",
"Huawei",
""
],
[
"Cheng",
"Xueqi",
""
]
] |
2505.17702 | Xueyang Li | Xueyang Li, Jiahao Li, Yu Song, Yunzhong Lou, Xiangdong Zhou | Seek-CAD: A Self-refined Generative Modeling for 3D Parametric CAD Using Local Inference via DeepSeek | null | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The advent of Computer-Aided Design (CAD) generative modeling will significantly transform the design of industrial products. The recent research endeavor has extended into the realm of Large Language Models (LLMs). In contrast to fine-tuning methods, training-free approaches typically utilize the advanced closed-source LLMs, thereby offering enhanced flexibility and efficiency in the development of AI agents for generating CAD parametric models. However, the substantial cost and limitations of local deployment of the top-tier closed-source LLMs pose challenges in practical applications. The Seek-CAD is the pioneer exploration of locally deployed open-source inference LLM DeepSeek-R1 for CAD parametric model generation with a training-free methodology. This study is the first investigation to incorporate both visual and Chain-of-Thought (CoT) feedback within the self-refinement mechanism for generating CAD models. Specifically, the initial generated parametric CAD model is rendered into a sequence of step-wise perspective images, which are subsequently processed by a Vision Language Model (VLM) alongside the corresponding CoTs derived from DeepSeek-R1 to assess the CAD model generation. Then, the feedback is utilized by DeepSeek-R1 to refine the initial generated model for the next round of generation. Moreover, we present an innovative 3D CAD model dataset structured around the SSR (Sketch, Sketch-based feature, and Refinements) triple design paradigm. This dataset encompasses a wide range of CAD commands, thereby aligning effectively with industrial application requirements and proving suitable for the generation of LLMs. Extensive experiments validate the effectiveness of Seek-CAD under various metrics. | [
{
"version": "v1",
"created": "Fri, 23 May 2025 10:11:19 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 07:35:10 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Li",
"Xueyang",
""
],
[
"Li",
"Jiahao",
""
],
[
"Song",
"Yu",
""
],
[
"Lou",
"Yunzhong",
""
],
[
"Zhou",
"Xiangdong",
""
]
] |
2505.17955 | Yujin Jeong | Yujin Jeong, Arnas Uselis, Seong Joon Oh, Anna Rohrbach | Diffusion Classifiers Understand Compositionality, but Conditions Apply | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Understanding visual scenes is fundamental to human intelligence. While discriminative models have significantly advanced computer vision, they often struggle with compositional understanding. In contrast, recent generative text-to-image diffusion models excel at synthesizing complex scenes, suggesting inherent compositional capabilities. Building on this, zero-shot diffusion classifiers have been proposed to repurpose diffusion models for discriminative tasks. While prior work offered promising results in discriminative compositional scenarios, these results remain preliminary due to a small number of benchmarks and a relatively shallow analysis of conditions under which the models succeed. To address this, we present a comprehensive study of the discriminative capabilities of diffusion classifiers on a wide range of compositional tasks. Specifically, our study covers three diffusion models (SD 1.5, 2.0, and, for the first time, 3-m) spanning 10 datasets and over 30 tasks. Further, we shed light on the role that target dataset domains play in respective performance; to isolate the domain effects, we introduce a new diagnostic benchmark Self-Bench comprised of images created by diffusion models themselves. Finally, we explore the importance of timestep weighting and uncover a relationship between domain gap and timestep sensitivity, particularly for SD3-m. To sum up, diffusion classifiers understand compositionality, but conditions apply! Code and dataset are available at https://github.com/eugene6923/Diffusion-Classifiers-Compositionality. | [
{
"version": "v1",
"created": "Fri, 23 May 2025 14:29:52 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 17:59:50 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Jeong",
"Yujin",
""
],
[
"Uselis",
"Arnas",
""
],
[
"Oh",
"Seong Joon",
""
],
[
"Rohrbach",
"Anna",
""
]
] |
2505.18128 | Chau Minh Pham | Chau Minh Pham and Jenna Russell and Dzung Pham and Mohit Iyyer | Frankentext: Stitching random text fragments into long-form narratives | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | We introduce Frankentexts, a new type of long-form narratives produced by LLMs under the extreme constraint that most tokens (e.g., 90%) must be copied verbatim from human writings. This task presents a challenging test of controllable generation, requiring models to satisfy a writing prompt, integrate disparate text fragments, and still produce a coherent narrative. To generate Frankentexts, we instruct the model to produce a draft by selecting and combining human-written passages, then iteratively revise the draft while maintaining a user-specified copy ratio. We evaluate the resulting Frankentexts along three axes: writing quality, instruction adherence, and detectability. Gemini-2.5-Pro performs surprisingly well on this task: 81% of its Frankentexts are coherent and 100% relevant to the prompt. Notably, up to 59% of these outputs are misclassified as human-written by detectors like Pangram, revealing limitations in AI text detectors. Human annotators can sometimes identify Frankentexts through their abrupt tone shifts and inconsistent grammar between segments, especially in longer generations. Beyond presenting a challenging generation task, Frankentexts invite discussion on building effective detectors for this new grey zone of authorship, provide training data for mixed authorship detection, and serve as a sandbox for studying human-AI co-writing processes. | [
{
"version": "v1",
"created": "Fri, 23 May 2025 17:38:47 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 01:43:46 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Pham",
"Chau Minh",
""
],
[
"Russell",
"Jenna",
""
],
[
"Pham",
"Dzung",
""
],
[
"Iyyer",
"Mohit",
""
]
] |
2505.18133 | Nitin Jha | Nitin Jha, Abhishek Parakh, Mahadevan Subramaniam | Joint Encryption and Error Correction for Secure Quantum Communication | null | Sci Rep 14, 24542 (2024) | 10.1038/s41598-024-75212-8 | null | quant-ph cs.NI | http://creativecommons.org/licenses/by/4.0/ | Secure quantum networks are a bedrock requirement for developing a future quantum internet. However, quantum channels are susceptible to channel noise that introduce errors in the transmitted data. The traditional approach to providing error correction typically encapsulates the message in an error correction code after encryption. Such separate processes incur overhead that must be avoided when possible. We, consequently, provide a single integrated process that allows for encryption as well as error correction. This is a first attempt to do so for secure quantum communication and combines the Calderbank-Shor-Steane (CSS) code with the three-stage secure quantum communication protocol. Lastly, it allows for arbitrary qubits to be transmitted from sender to receiver making the proposed protocol general purpose. | [
{
"version": "v1",
"created": "Fri, 23 May 2025 17:42:03 GMT"
},
{
"version": "v2",
"created": "Mon, 26 May 2025 02:25:33 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Jha",
"Nitin",
""
],
[
"Parakh",
"Abhishek",
""
],
[
"Subramaniam",
"Mahadevan",
""
]
] |
2505.18325 | Licheng Pan | Licheng Pan, Yongqi Tong, Xin Zhang, Xiaolu Zhang, Jun Zhou, Zhixuan Chu | Understanding and Mitigating Overrefusal in LLMs from an Unveiling Perspective of Safety Decision Boundary | We have identified significant errors in the results presented in this paper, specifically in the evaluation sections concerning the DPO training of LLaMA2 and Qwen2.5, as well as in the representation space visualization section. Given the extent of these issues, we intend to substantially revise the manuscript's content and structure. Hence, we request to withdraw it from arXiv at this time | null | null | null | cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Large language models (LLMs) have demonstrated remarkable capabilities across a wide range of tasks, yet they often refuse to answer legitimate queries-a phenomenon known as overrefusal. Overrefusal typically stems from over-conservative safety alignment, causing models to treat many reasonable prompts as potentially risky. To systematically understand this issue, we probe and leverage the models'safety decision boundaries to analyze and mitigate overrefusal. Our findings reveal that overrefusal is closely tied to misalignment at these boundary regions, where models struggle to distinguish subtle differences between benign and harmful content. Building on these insights, we present RASS, an automated framework for prompt generation and selection that strategically targets overrefusal prompts near the safety boundary. By harnessing steering vectors in the representation space, RASS efficiently identifies and curates boundary-aligned prompts, enabling more effective and targeted mitigation of overrefusal. This approach not only provides a more precise and interpretable view of model safety decisions but also seamlessly extends to multilingual scenarios.We have explored the safety decision boundaries of various LLMs and construct the MORBench evaluation set to facilitate robust assessment of model safety and helpfulness across multiple languages. Code and datasets will be released at https://anonymous.4open.science/r/RASS-80D3. | [
{
"version": "v1",
"created": "Fri, 23 May 2025 19:30:49 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 13:22:42 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Pan",
"Licheng",
""
],
[
"Tong",
"Yongqi",
""
],
[
"Zhang",
"Xin",
""
],
[
"Zhang",
"Xiaolu",
""
],
[
"Zhou",
"Jun",
""
],
[
"Chu",
"Zhixuan",
""
]
] |
2505.18329 | Sophie Libkind | Sophie Libkind and David Jaz Myers | Towards a double operadic theory of systems | null | null | null | null | math.CT | http://creativecommons.org/licenses/by/4.0/ | We present a unified framework for categorical systems theory which packages a collection of open systems, their interactions, and their maps into a symmetric monoidal loose right module of systems over a symmetric monoidal double category of interfaces and interactions. As examples, we give detailed descriptions of (1) the module of open Petri nets over undirected wiring diagrams and (2) the module of deterministic Moore machines over lenses. We define several pseudo-functorial constructions of modules of systems in the form of doctrines of systems theories. In particular, we introduce doctrines for port-plugging systems, variable sharing systems, and generalized Moore machines, each of which generalizes existing work in categorical systems theory. Finally, we observe how diagrammatic interaction patterns are free processes in particular doctrines. | [
{
"version": "v1",
"created": "Fri, 23 May 2025 19:35:57 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 19:20:26 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Libkind",
"Sophie",
""
],
[
"Myers",
"David Jaz",
""
]
] |
2505.18377 | Pingchuan Ma | Pingchuan Ma, Ziang Yin, Qi Jing, Zhengqi Gao, Nicholas Gangi, Boyang Zhang, Tsung-Wei Huang, Zhaoran Huang, Duane S. Boning, Yu Yao, and Jiaqi Gu | SP2RINT: Spatially-Decoupled Physics-Inspired Progressive Inverse Optimization for Scalable, PDE-Constrained Meta-Optical Neural Network Training | null | null | null | null | physics.optics cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | DONNs leverage light propagation for efficient analog AI and signal processing. Advances in nanophotonic fabrication and metasurface-based wavefront engineering have opened new pathways to realize high-capacity DONNs across various spectral regimes. Training such DONN systems to determine the metasurface structures remains challenging. Heuristic methods are fast but oversimplify metasurfaces modulation, often resulting in physically unrealizable designs and significant performance degradation. Simulation-in-the-loop optimizes implementable metasurfaces via adjoint methods, but is computationally prohibitive and unscalable. To address these limitations, we propose SP2RINT, a spatially decoupled, progressive training framework that formulates DONN training as a PDE-constrained learning problem. Metasurface responses are first relaxed into freely trainable transfer matrices with a banded structure. We then progressively enforce physical constraints by alternating between transfer matrix training and adjoint-based inverse design, avoiding per-iteration PDE solves while ensuring final physical realizability. To further reduce runtime, we introduce a physics-inspired, spatially decoupled inverse design strategy based on the natural locality of field interactions. This approach partitions the metasurface into independently solvable patches, enabling scalable and parallel inverse design with system-level calibration. Evaluated across diverse DONN training tasks, SP2RINT achieves digital-comparable accuracy while being 1825 times faster than simulation-in-the-loop approaches. By bridging the gap between abstract DONN models and implementable photonic hardware, SP2RINT enables scalable, high-performance training of physically realizable meta-optical neural systems. Our code is available at https://github.com/ScopeX-ASU/SP2RINT | [
{
"version": "v1",
"created": "Fri, 23 May 2025 21:05:40 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 22:12:11 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Ma",
"Pingchuan",
""
],
[
"Yin",
"Ziang",
""
],
[
"Jing",
"Qi",
""
],
[
"Gao",
"Zhengqi",
""
],
[
"Gangi",
"Nicholas",
""
],
[
"Zhang",
"Boyang",
""
],
[
"Huang",
"Tsung-Wei",
""
],
[
"Huang",
"Zhaoran",
""
],
[
"Boning",
"Duane S.",
""
],
[
"Yao",
"Yu",
""
],
[
"Gu",
"Jiaqi",
""
]
] |
2505.18422 | Saleh Afroogh | Saleh Afroogh, Kush R. Varshney, Jason DCruz | A Task-Driven Human-AI Collaboration: When to Automate, When to Collaborate, When to Challenge | null | null | null | null | cs.CY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | According to several empirical investigations, despite enhancing human capabilities, human-AI cooperation frequently falls short of expectations and fails to reach true synergy. We propose a task-driven framework that reverses prevalent approaches by assigning AI roles according to how the task's requirements align with the capabilities of AI technology. Three major AI roles are identified through task analysis across risk and complexity dimensions: autonomous, assistive/collaborative, and adversarial. We show how proper human-AI integration maintains meaningful agency while improving performance by methodically mapping these roles to various task types based on current empirical findings. This framework lays the foundation for practically effective and morally sound human-AI collaboration that unleashes human potential by aligning task attributes to AI capabilities. It also provides structured guidance for context-sensitive automation that complements human strengths rather than replacing human judgment. | [
{
"version": "v1",
"created": "Fri, 23 May 2025 23:19:15 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 02:30:11 GMT"
},
{
"version": "v3",
"created": "Wed, 28 May 2025 18:22:17 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Afroogh",
"Saleh",
""
],
[
"Varshney",
"Kush R.",
""
],
[
"DCruz",
"Jason",
""
]
] |
2505.18424 | Juampablo Heras Rivera | Tianyi Ren, Juampablo E. Heras Rivera, Hitender Oswal, Yutong Pan, William Henry, Sophie Walters, Mehmet Kurt | How We Won the ISLES'24 Challenge by Preprocessing | null | null | null | null | eess.IV cs.AI cs.CV | http://creativecommons.org/licenses/by/4.0/ | Stroke is among the top three causes of death worldwide, and accurate identification of stroke lesion boundaries is critical for diagnosis and treatment. Supervised deep learning methods have emerged as the leading solution for stroke lesion segmentation but require large, diverse, and annotated datasets. The ISLES'24 challenge addresses this need by providing longitudinal stroke imaging data, including CT scans taken on arrival to the hospital and follow-up MRI taken 2-9 days from initial arrival, with annotations derived from follow-up MRI. Importantly, models submitted to the ISLES'24 challenge are evaluated using only CT inputs, requiring prediction of lesion progression that may not be visible in CT scans for segmentation. Our winning solution shows that a carefully designed preprocessing pipeline including deep-learning-based skull stripping and custom intensity windowing is beneficial for accurate segmentation. Combined with a standard large residual nnU-Net architecture for segmentation, this approach achieves a mean test Dice of 28.5 with a standard deviation of 21.27. | [
{
"version": "v1",
"created": "Fri, 23 May 2025 23:25:00 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 22:02:21 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Ren",
"Tianyi",
""
],
[
"Rivera",
"Juampablo E. Heras",
""
],
[
"Oswal",
"Hitender",
""
],
[
"Pan",
"Yutong",
""
],
[
"Henry",
"William",
""
],
[
"Walters",
"Sophie",
""
],
[
"Kurt",
"Mehmet",
""
]
] |
2505.18447 | Yue Wang | Chi Zhang, Ziying Jia, George K. Atia, Sihong He, Yue Wang | Pessimism Principle Can Be Effective: Towards a Framework for Zero-Shot Transfer Reinforcement Learning | Accepted to ICML 2025 | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Transfer reinforcement learning aims to derive a near-optimal policy for a target environment with limited data by leveraging abundant data from related source domains. However, it faces two key challenges: the lack of performance guarantees for the transferred policy, which can lead to undesired actions, and the risk of negative transfer when multiple source domains are involved. We propose a novel framework based on the pessimism principle, which constructs and optimizes a conservative estimation of the target domain's performance. Our framework effectively addresses the two challenges by providing an optimized lower bound on target performance, ensuring safe and reliable decisions, and by exhibiting monotonic improvement with respect to the quality of the source domains, thereby avoiding negative transfer. We construct two types of conservative estimations, rigorously characterize their effectiveness, and develop efficient distributed algorithms with convergence guarantees. Our framework provides a theoretically sound and practically robust solution for transfer learning in reinforcement learning. | [
{
"version": "v1",
"created": "Sat, 24 May 2025 01:09:10 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 14:15:50 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhang",
"Chi",
""
],
[
"Jia",
"Ziying",
""
],
[
"Atia",
"George K.",
""
],
[
"He",
"Sihong",
""
],
[
"Wang",
"Yue",
""
]
] |
2505.18459 | Viacheslav Tsaran | Viacheslav Tsaran, Francesco Marino, Sonia Bacca, Francesca Bonaiti, Marc Vanderhaeghen | Bridging reaction theory and nuclear structure in $\pi^\pm$-${}^{48}$Ca scattering | null | null | null | MITP-25-039 | nucl-th hep-ph | http://creativecommons.org/licenses/by/4.0/ | We extend the pion-nucleus multiple-scattering framework to include detailed second-order rescattering dynamics for nuclei with non-zero isospin. To account for intermediate charge-exchange and nucleon spin-flip effects, we develop a scattering potential that depends on the one- and two-body densities of the target nucleus. We compute one-body densities from coupled-cluster theory and two-body densities within the Hartree-Fock approximation. To estimate theoretical uncertainties, we employ modern nuclear Hamiltonians derived from chiral effective field theory. While the sensitivity to nuclear structure details is mild, second-order corrections are found to be sizeable and essential for accurately reproducing differential cross sections measured in $\pi^\pm$-${}^{48}$Ca elastic scattering within the $\Delta(1232)$-resonance region | [
{
"version": "v1",
"created": "Sat, 24 May 2025 01:57:45 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Tsaran",
"Viacheslav",
""
],
[
"Marino",
"Francesco",
""
],
[
"Bacca",
"Sonia",
""
],
[
"Bonaiti",
"Francesca",
""
],
[
"Vanderhaeghen",
"Marc",
""
]
] |
2505.18477 | Fukun Liu | Fukun Liu, Adam T. Greer, Gengchen Mai, Jin Sun | ZooplanktonBench: A Geo-Aware Zooplankton Recognition and Classification Dataset from Marine Observations | Accepted to KDD 2025 Datasets and Benchmarks Track | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Plankton are small drifting organisms found throughout the world's oceans and can be indicators of ocean health. One component of this plankton community is the zooplankton, which includes gelatinous animals and crustaceans (e.g. shrimp), as well as the early life stages (i.e., eggs and larvae) of many commercially important fishes. Being able to monitor zooplankton abundances accurately and understand how populations change in relation to ocean conditions is invaluable to marine science research, with important implications for future marine seafood productivity. While new imaging technologies generate massive amounts of video data of zooplankton, analyzing them using general-purpose computer vision tools turns out to be highly challenging due to the high similarity in appearance between the zooplankton and its background (e.g., marine snow). In this work, we present the ZooplanktonBench, a benchmark dataset containing images and videos of zooplankton associated with rich geospatial metadata (e.g., geographic coordinates, depth, etc.) in various water ecosystems. ZooplanktonBench defines a collection of tasks to detect, classify, and track zooplankton in challenging settings, including highly cluttered environments, living vs non-living classification, objects with similar shapes, and relatively small objects. Our dataset presents unique challenges and opportunities for state-of-the-art computer vision systems to evolve and improve visual understanding in dynamic environments characterized by significant variation and the need for geo-awareness. The code and settings described in this paper can be found on our website: https://lfk118.github.io/ZooplanktonBench_Webpage. | [
{
"version": "v1",
"created": "Sat, 24 May 2025 02:49:52 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 20:23:19 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 01:01:52 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Liu",
"Fukun",
""
],
[
"Greer",
"Adam T.",
""
],
[
"Mai",
"Gengchen",
""
],
[
"Sun",
"Jin",
""
]
] |
2505.18486 | Hong Jiao | Hong Jiao, Dan Song, Won-Chan Lee | Comparing Human and AI Rater Effects Using the Many-Facet Rasch Model | null | null | null | null | cs.CL cs.LG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Large language models (LLMs) have been widely explored for automated scoring in low-stakes assessment to facilitate learning and instruction. Empirical evidence related to which LLM produces the most reliable scores and induces least rater effects needs to be collected before the use of LLMs for automated scoring in practice. This study compared ten LLMs (ChatGPT 3.5, ChatGPT 4, ChatGPT 4o, OpenAI o1, Claude 3.5 Sonnet, Gemini 1.5, Gemini 1.5 Pro, Gemini 2.0, as well as DeepSeek V3, and DeepSeek R1) with human expert raters in scoring two types of writing tasks. The accuracy of the holistic and analytic scores from LLMs compared with human raters was evaluated in terms of Quadratic Weighted Kappa. Intra-rater consistency across prompts was compared in terms of Cronbach Alpha. Rater effects of LLMs were evaluated and compared with human raters using the Many-Facet Rasch model. The results in general supported the use of ChatGPT 4o, Gemini 1.5 Pro, and Claude 3.5 Sonnet with high scoring accuracy, better rater reliability, and less rater effects. | [
{
"version": "v1",
"created": "Sat, 24 May 2025 03:24:44 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 20:07:32 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Jiao",
"Hong",
""
],
[
"Song",
"Dan",
""
],
[
"Lee",
"Won-Chan",
""
]
] |
2505.18542 | Chen Yang | Chen Yang, Ruping Xu, Ruizhe Li, Bin Cao, Jing Fan | Business as Rulesual: A Benchmark and Framework for Business Rule Flow Modeling with LLMs | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Process mining aims to discover, monitor and optimize the actual behaviors of real processes. While prior work has mainly focused on extracting procedural action flows from instructional texts, rule flows embedded in business documents remain underexplored. To this end, we introduce a novel annotated Chinese dataset, BPRF, which contains 50 business process documents with 326 explicitly labeled business rules across multiple domains. Each rule is represented as a <Condition, Action> pair, and we annotate logical dependencies between rules (sequential, conditional, or parallel). We also propose ExIde, a framework for automatic business rule extraction and dependency relationship identification using large language models (LLMs). We evaluate ExIde using 12 state-of-the-art (SOTA) LLMs on the BPRF dataset, benchmarking performance on both rule extraction and dependency classification tasks of current LLMs. Our results demonstrate the effectiveness of ExIde in extracting structured business rules and analyzing their interdependencies for current SOTA LLMs, paving the way for more automated and interpretable business process automation. | [
{
"version": "v1",
"created": "Sat, 24 May 2025 06:13:35 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 01:22:02 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Yang",
"Chen",
""
],
[
"Xu",
"Ruping",
""
],
[
"Li",
"Ruizhe",
""
],
[
"Cao",
"Bin",
""
],
[
"Fan",
"Jing",
""
]
] |
2505.18638 | Md. Tanzib Hosain | Md. Tanzib Hosain, Rajan Das Gupta, Md. Kishor Morol | Multilingual Question Answering in Low-Resource Settings: A Dzongkha-English Benchmark for Foundation Models | 24 pages, 20 figures | null | null | null | cs.CL | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In this work, we provide DZEN, a dataset of parallel Dzongkha and English test questions for Bhutanese middle and high school students. The over 5K questions in our collection span a variety of scientific topics and include factual, application, and reasoning-based questions. We use our parallel dataset to test a number of Large Language Models (LLMs) and find a significant performance difference between the models in English and Dzongkha. We also look at different prompting strategies and discover that Chain-of-Thought (CoT) prompting works well for reasoning questions but less well for factual ones. We also find that adding English translations enhances the precision of Dzongkha question responses. Our results point to exciting avenues for further study to improve LLM performance in Dzongkha and, more generally, in low-resource languages. We release the dataset at: https://github.com/kraritt/llm_dzongkha_evaluation. | [
{
"version": "v1",
"created": "Sat, 24 May 2025 11:01:05 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 17:11:54 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Hosain",
"Md. Tanzib",
""
],
[
"Gupta",
"Rajan Das",
""
],
[
"Morol",
"Md. Kishor",
""
]
] |
2505.18686 | Silin Cheng | Yang Liu, Silin Cheng, Xinwei He, Sebastien Ourselin, Lei Tan, Gen Luo | WeakMCN: Multi-task Collaborative Network for Weakly Supervised Referring Expression Comprehension and Segmentation | Accepted by CVPR2025 | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Weakly supervised referring expression comprehension(WREC) and segmentation(WRES) aim to learn object grounding based on a given expression using weak supervision signals like image-text pairs. While these tasks have traditionally been modeled separately, we argue that they can benefit from joint learning in a multi-task framework. To this end, we propose WeakMCN, a novel multi-task collaborative network that effectively combines WREC and WRES with a dual-branch architecture. Specifically, the WREC branch is formulated as anchor-based contrastive learning, which also acts as a teacher to supervise the WRES branch. In WeakMCN, we propose two innovative designs to facilitate multi-task collaboration, namely Dynamic Visual Feature Enhancement(DVFE) and Collaborative Consistency Module(CCM). DVFE dynamically combines various pre-trained visual knowledge to meet different task requirements, while CCM promotes cross-task consistency from the perspective of optimization. Extensive experimental results on three popular REC and RES benchmarks, i.e., RefCOCO, RefCOCO+, and RefCOCOg, consistently demonstrate performance gains of WeakMCN over state-of-the-art single-task alternatives, e.g., up to 3.91% and 13.11% on RefCOCO for WREC and WRES tasks, respectively. Furthermore, experiments also validate the strong generalization ability of WeakMCN in both semi-supervised REC and RES settings against existing methods, e.g., +8.94% for semi-REC and +7.71% for semi-RES on 1% RefCOCO. The code is publicly available at https://github.com/MRUIL/WeakMCN. | [
{
"version": "v1",
"created": "Sat, 24 May 2025 13:05:17 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 00:58:26 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Liu",
"Yang",
""
],
[
"Cheng",
"Silin",
""
],
[
"He",
"Xinwei",
""
],
[
"Ourselin",
"Sebastien",
""
],
[
"Tan",
"Lei",
""
],
[
"Luo",
"Gen",
""
]
] |
2505.18893 | Reva Schwartz | Reva Schwartz, Rumman Chowdhury, Akash Kundu, Heather Frase, Marzieh Fadaee, Tom David, Gabriella Waters, Afaf Taik, Morgan Briggs, Patrick Hall, Shomik Jain, Kyra Yee, Spencer Thomas, Sundeep Bhandari, Qinghua Lu, Matthew Holmes, Theodora Skeadas | Reality Check: A New Evaluation Ecosystem Is Necessary to Understand AI's Real World Effects | 9 pages | null | null | null | cs.CY cs.AI | http://creativecommons.org/licenses/by/4.0/ | Conventional AI evaluation approaches concentrated within the AI stack exhibit systemic limitations for exploring, navigating and resolving the human and societal factors that play out in real world deployment such as in education, finance, healthcare, and employment sectors. AI capability evaluations can capture detail about first-order effects, such as whether immediate system outputs are accurate, or contain toxic, biased or stereotypical content, but AI's second-order effects, i.e. any long-term outcomes and consequences that may result from AI use in the real world, have become a significant area of interest as the technology becomes embedded in our daily lives. These secondary effects can include shifts in user behavior, societal, cultural and economic ramifications, workforce transformations, and long-term downstream impacts that may result from a broad and growing set of risks. This position paper argues that measuring the indirect and secondary effects of AI will require expansion beyond static, single-turn approaches conducted in silico to include testing paradigms that can capture what actually materializes when people use AI technology in context. Specifically, we describe the need for data and methods that can facilitate contextual awareness and enable downstream interpretation and decision making about AI's secondary effects, and recommend requirements for a new ecosystem. | [
{
"version": "v1",
"created": "Sat, 24 May 2025 22:35:32 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 13:26:32 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 13:42:10 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Schwartz",
"Reva",
""
],
[
"Chowdhury",
"Rumman",
""
],
[
"Kundu",
"Akash",
""
],
[
"Frase",
"Heather",
""
],
[
"Fadaee",
"Marzieh",
""
],
[
"David",
"Tom",
""
],
[
"Waters",
"Gabriella",
""
],
[
"Taik",
"Afaf",
""
],
[
"Briggs",
"Morgan",
""
],
[
"Hall",
"Patrick",
""
],
[
"Jain",
"Shomik",
""
],
[
"Yee",
"Kyra",
""
],
[
"Thomas",
"Spencer",
""
],
[
"Bhandari",
"Sundeep",
""
],
[
"Lu",
"Qinghua",
""
],
[
"Holmes",
"Matthew",
""
],
[
"Skeadas",
"Theodora",
""
]
] |
2505.18962 | Xiaoqiang Wang | Xiaoqiang Wang, Suyuchen Wang, Yun Zhu, Bang Liu | System-1.5 Reasoning: Traversal in Language and Latent Spaces with Dynamic Shortcuts | Work in progress | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Chain-of-thought (CoT) reasoning enables large language models (LLMs) to move beyond fast System-1 responses and engage in deliberative System-2 reasoning. However, this comes at the cost of significant inefficiency due to verbose intermediate output. Recent latent-space reasoning methods improve efficiency by operating on hidden states without decoding into language, yet they treat all steps uniformly, failing to distinguish critical deductions from auxiliary steps and resulting in suboptimal use of computational resources. In this paper, we propose System-1.5 Reasoning, an adaptive reasoning framework that dynamically allocates computation across reasoning steps through shortcut paths in latent space. Specifically, System-1.5 Reasoning introduces two types of dynamic shortcuts. The model depth shortcut (DS) adaptively reasons along the vertical depth by early exiting non-critical tokens through lightweight adapter branches, while allowing critical tokens to continue through deeper Transformer layers. The step shortcut (SS) reuses hidden states across the decoding steps to skip trivial steps and reason horizontally in latent space. Training System-1.5 Reasoning involves a two-stage self-distillation process: first distilling natural language CoT into latent-space continuous thought, and then distilling full-path System-2 latent reasoning into adaptive shortcut paths (System-1.5 Reasoning). Experiments on reasoning tasks demonstrate the superior performance of our method. For example, on GSM8K, System-1.5 Reasoning achieves reasoning performance comparable to traditional CoT fine-tuning methods while accelerating inference by over 20x and reducing token generation by 92.31% on average. | [
{
"version": "v1",
"created": "Sun, 25 May 2025 03:35:49 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 07:35:48 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wang",
"Xiaoqiang",
""
],
[
"Wang",
"Suyuchen",
""
],
[
"Zhu",
"Yun",
""
],
[
"Liu",
"Bang",
""
]
] |
2505.19164 | Soumik Dey | Ashirbad Mishra, Jinyu Zhao, Soumik Dey, Hansi Wu, Binbin Li, Kamesh Madduri | BroadGen: A Framework for Generating Effective and Efficient Advertiser Broad Match Keyphrase Recommendations | null | null | null | null | cs.IR cs.AI cs.LG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In the domain of sponsored search advertising, the focus of Keyphrase recommendation has largely been on exact match types, which pose issues such as high management expenses, limited targeting scope, and evolving search query patterns. Alternatives like Broad match types can alleviate certain drawbacks of exact matches but present challenges like poor targeting accuracy and minimal supervisory signals owing to limited advertiser usage. This research defines the criteria for an ideal broad match, emphasizing on both efficiency and effectiveness, ensuring that a significant portion of matched queries are relevant. We propose BroadGen, an innovative framework that recommends efficient and effective broad match keyphrases by utilizing historical search query data. Additionally, we demonstrate that BroadGen, through token correspondence modeling, maintains better query stability over time. BroadGen's capabilities allow it to serve daily, millions of sellers at eBay with over 2.3 billion items. | [
{
"version": "v1",
"created": "Sun, 25 May 2025 14:25:52 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 05:45:19 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Mishra",
"Ashirbad",
""
],
[
"Zhao",
"Jinyu",
""
],
[
"Dey",
"Soumik",
""
],
[
"Wu",
"Hansi",
""
],
[
"Li",
"Binbin",
""
],
[
"Madduri",
"Kamesh",
""
]
] |
2505.19201 | Yunhai Hu Mr. | Yunhai Hu, Tianhua Xia, Zining Liu, Rahul Raman, Xingyu Liu, Bo Bao, Eric Sather, Vithursan Thangarasa, and Sai Qian Zhang | DREAM: Drafting with Refined Target Features and Entropy-Adaptive Cross-Attention Fusion for Multimodal Speculative Decoding | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Speculative decoding (SD) has emerged as a powerful method for accelerating autoregressive generation in large language models (LLMs), yet its integration into vision-language models (VLMs) remains underexplored. We introduce DREAM, a novel speculative decoding framework tailored for VLMs that combines three key innovations: (1) a cross-attention-based mechanism to inject intermediate features from the target model into the draft model for improved alignment, (2) adaptive intermediate feature selection based on attention entropy to guide efficient draft model training, and (3) visual token compression to reduce draft model latency. DREAM enables efficient, accurate, and parallel multimodal decoding with significant throughput improvement. Experiments across a diverse set of recent popular VLMs, including LLaVA, Pixtral, SmolVLM and Gemma3, demonstrate up to 3.6x speedup over conventional decoding and significantly outperform prior SD baselines in both inference throughput and speculative draft acceptance length across a broad range of multimodal benchmarks. The code is publicly available at: https://github.com/SAI-Lab-NYU/DREAM.git | [
{
"version": "v1",
"created": "Sun, 25 May 2025 15:56:50 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 12:40:23 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Hu",
"Yunhai",
""
],
[
"Xia",
"Tianhua",
""
],
[
"Liu",
"Zining",
""
],
[
"Raman",
"Rahul",
""
],
[
"Liu",
"Xingyu",
""
],
[
"Bao",
"Bo",
""
],
[
"Sather",
"Eric",
""
],
[
"Thangarasa",
"Vithursan",
""
],
[
"Zhang",
"Sai Qian",
""
]
] |
2505.19256 | Vivek Gopalakrishnan | Vivek Gopalakrishnan, Neel Dey, Polina Golland | PolyPose: Localizing Deformable Anatomy in 3D from Sparse 2D X-ray Images using Polyrigid Transforms | Code available at https://github.com/eigenvivek/polypose | null | null | null | cs.CV physics.med-ph | http://creativecommons.org/licenses/by/4.0/ | Determining the 3D pose of a patient from a limited set of 2D X-ray images is a critical task in interventional settings. While preoperative volumetric imaging (e.g., CT and MRI) provides precise 3D localization and visualization of anatomical targets, these modalities cannot be acquired during procedures, where fast 2D imaging (X-ray) is used instead. To integrate volumetric guidance into intraoperative procedures, we present PolyPose, a simple and robust method for deformable 2D/3D registration. PolyPose parameterizes complex 3D deformation fields as a composition of rigid transforms, leveraging the biological constraint that individual bones do not bend in typical motion. Unlike existing methods that either assume no inter-joint movement or fail outright in this under-determined setting, our polyrigid formulation enforces anatomically plausible priors that respect the piecewise rigid nature of human movement. This approach eliminates the need for expensive deformation regularizers that require patient- and procedure-specific hyperparameter optimization. Across extensive experiments on diverse datasets from orthopedic surgery and radiotherapy, we show that this strong inductive bias enables PolyPose to successfully align the patient's preoperative volume to as few as two X-ray images, thereby providing crucial 3D guidance in challenging sparse-view and limited-angle settings where current registration methods fail. | [
{
"version": "v1",
"created": "Sun, 25 May 2025 18:24:18 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 21:02:56 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Gopalakrishnan",
"Vivek",
""
],
[
"Dey",
"Neel",
""
],
[
"Golland",
"Polina",
""
]
] |
2505.19301 | Vineeth Sai Narajala | Ken Huang, Vineeth Sai Narajala, John Yeoh, Jason Ross, Ramesh Raskar, Youssef Harkati, Jerry Huang, Idan Habler, Chris Hughes | A Novel Zero-Trust Identity Framework for Agentic AI: Decentralized Authentication and Fine-Grained Access Control | 24 Pages, 5 figures, 2 tables, edit: added a missed Author | null | null | null | cs.CR cs.AI cs.MA | http://creativecommons.org/licenses/by/4.0/ | Traditional Identity and Access Management (IAM) systems, primarily designed for human users or static machine identities via protocols such as OAuth, OpenID Connect (OIDC), and SAML, prove fundamentally inadequate for the dynamic, interdependent, and often ephemeral nature of AI agents operating at scale within Multi Agent Systems (MAS), a computational system composed of multiple interacting intelligent agents that work collectively.
This paper posits the imperative for a novel Agentic AI IAM framework: We deconstruct the limitations of existing protocols when applied to MAS, illustrating with concrete examples why their coarse-grained controls, single-entity focus, and lack of context-awareness falter. We then propose a comprehensive framework built upon rich, verifiable Agent Identities (IDs), leveraging Decentralized Identifiers (DIDs) and Verifiable Credentials (VCs), that encapsulate an agents capabilities, provenance, behavioral scope, and security posture.
Our framework includes an Agent Naming Service (ANS) for secure and capability-aware discovery, dynamic fine-grained access control mechanisms, and critically, a unified global session management and policy enforcement layer for real-time control and consistent revocation across heterogeneous agent communication protocols. We also explore how Zero-Knowledge Proofs (ZKPs) enable privacy-preserving attribute disclosure and verifiable policy compliance.
We outline the architecture, operational lifecycle, innovative contributions, and security considerations of this new IAM paradigm, aiming to establish the foundational trust, accountability, and security necessary for the burgeoning field of agentic AI and the complex ecosystems they will inhabit. | [
{
"version": "v1",
"created": "Sun, 25 May 2025 20:21:55 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 17:37:25 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Huang",
"Ken",
""
],
[
"Narajala",
"Vineeth Sai",
""
],
[
"Yeoh",
"John",
""
],
[
"Ross",
"Jason",
""
],
[
"Raskar",
"Ramesh",
""
],
[
"Harkati",
"Youssef",
""
],
[
"Huang",
"Jerry",
""
],
[
"Habler",
"Idan",
""
],
[
"Hughes",
"Chris",
""
]
] |
2505.19312 | Zirui Li | Zirui Li, Siwei Wu, Xingyu Wang, Yi Zhou, Yizhi Li, Chenghua Lin | DocMMIR: A Framework for Document Multi-modal Information Retrieval | 13 pages, 7 figures. Code and data publicly available at https://github.com/J1mL1/DocMMIR | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The rapid advancement of unsupervised representation learning and large-scale pre-trained vision-language models has significantly improved cross-modal retrieval tasks. However, existing multi-modal information retrieval (MMIR) studies lack a comprehensive exploration of document-level retrieval and suffer from the absence of cross-domain datasets at this granularity. To address this limitation, we introduce DocMMIR, a novel multi-modal document retrieval framework designed explicitly to unify diverse document formats and domains, including Wikipedia articles, scientific papers (arXiv), and presentation slides, within a comprehensive retrieval scenario. We construct a large-scale cross-domain multimodal benchmark, comprising 450K samples, which systematically integrates textual and visual information. Our comprehensive experimental analysis reveals substantial limitations in current state-of-the-art MLLMs (CLIP, BLIP2, SigLIP-2, ALIGN) when applied to our tasks, with only CLIP demonstrating reasonable zero-shot performance. Furthermore, we conduct a systematic investigation of training strategies, including cross-modal fusion methods and loss functions, and develop a tailored approach to train CLIP on our benchmark. This results in a +31% improvement in MRR@10 compared to the zero-shot baseline. All our data and code are released in https://github.com/J1mL1/DocMMIR. | [
{
"version": "v1",
"created": "Sun, 25 May 2025 20:58:58 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 13:14:43 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Li",
"Zirui",
""
],
[
"Wu",
"Siwei",
""
],
[
"Wang",
"Xingyu",
""
],
[
"Zhou",
"Yi",
""
],
[
"Li",
"Yizhi",
""
],
[
"Lin",
"Chenghua",
""
]
] |
2505.19328 | Soufiane Belharbi | Manuela Gonz\'alez-Gonz\'alez, Soufiane Belharbi, Muhammad Osama Zeeshan, Masoumeh Sharafi, Muhammad Haseeb Aslam, Marco Pedersoli, Alessandro Lameiras Koerich, Simon L Bacon, Eric Granger | BAH Dataset for Ambivalence/Hesitancy Recognition in Videos for Behavioural Change | 41 pages, 13 figures, under review | null | null | null | cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | Recognizing complex emotions linked to ambivalence and hesitancy (A/H) can play a critical role in the personalization and effectiveness of digital behaviour change interventions. These subtle and conflicting emotions are manifested by a discord between multiple modalities, such as facial and vocal expressions, and body language. Although experts can be trained to identify A/H, integrating them into digital interventions is costly and less effective. Automatic learning systems provide a cost-effective alternative that can adapt to individual users, and operate seamlessly within real-time, and resource-limited environments. However, there are currently no datasets available for the design of ML models to recognize A/H. This paper introduces a first Behavioural Ambivalence/Hesitancy (BAH) dataset collected for subject-based multimodal recognition of A/H in videos. It contains videos from 224 participants captured across 9 provinces in Canada, with different age, and ethnicity. Through our web platform, we recruited participants to answer 7 questions, some of which were designed to elicit A/H while recording themselves via webcam with microphone. BAH amounts to 1,118 videos for a total duration of 8.26 hours with 1.5 hours of A/H. Our behavioural team annotated timestamp segments to indicate where A/H occurs, and provide frame- and video-level annotations with the A/H cues. Video transcripts and their timestamps are also included, along with cropped and aligned faces in each frame, and a variety of participants meta-data. We include results baselines for BAH at frame- and video-level recognition in multi-modal setups, in addition to zero-shot prediction, and for personalization using unsupervised domain adaptation. The limited performance of baseline models highlights the challenges of recognizing A/H in real-world videos. The data, code, and pretrained weights are available. | [
{
"version": "v1",
"created": "Sun, 25 May 2025 21:29:00 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 11:19:26 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"González-González",
"Manuela",
""
],
[
"Belharbi",
"Soufiane",
""
],
[
"Zeeshan",
"Muhammad Osama",
""
],
[
"Sharafi",
"Masoumeh",
""
],
[
"Aslam",
"Muhammad Haseeb",
""
],
[
"Pedersoli",
"Marco",
""
],
[
"Koerich",
"Alessandro Lameiras",
""
],
[
"Bacon",
"Simon L",
""
],
[
"Granger",
"Eric",
""
]
] |
2505.19398 | Ping Liu | Yiwei Xie and Ping Liu and Zheng Zhang | Erasing Concepts, Steering Generations: A Comprehensive Survey of Concept Suppression | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Text-to-Image (T2I) models have demonstrated impressive capabilities in generating high-quality and diverse visual content from natural language prompts. However, uncontrolled reproduction of sensitive, copyrighted, or harmful imagery poses serious ethical, legal, and safety challenges. To address these concerns, the concept erasure paradigm has emerged as a promising direction, enabling the selective removal of specific semantic concepts from generative models while preserving their overall utility. This survey provides a comprehensive overview and in-depth synthesis of concept erasure techniques in T2I diffusion models. We systematically categorize existing approaches along three key dimensions: intervention level, which identifies specific model components targeted for concept removal; optimization structure, referring to the algorithmic strategies employed to achieve suppression; and semantic scope, concerning the complexity and nature of the concepts addressed. This multi-dimensional taxonomy enables clear, structured comparisons across diverse methodologies, highlighting fundamental trade-offs between erasure specificity, generalization, and computational complexity. We further discuss current evaluation benchmarks, standardized metrics, and practical datasets, emphasizing gaps that limit comprehensive assessment, particularly regarding robustness and practical effectiveness. Finally, we outline major challenges and promising future directions, including disentanglement of concept representations, adaptive and incremental erasure strategies, adversarial robustness, and new generative architectures. This survey aims to guide researchers toward safer, more ethically aligned generative models, providing foundational knowledge and actionable recommendations to advance responsible development in generative AI. | [
{
"version": "v1",
"created": "Mon, 26 May 2025 01:24:34 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 14:48:02 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Xie",
"Yiwei",
""
],
[
"Liu",
"Ping",
""
],
[
"Zhang",
"Zheng",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.