Computer Vision
Auditing VLM Benchmarks from Existing Evaluation Runs
Denis Degtiarev, Anna Gromova, Sergey Katkov et al. (+5 more)
Every evaluation run of a
vision-
language model (VLM) already produces per-question artifacts (predictions, judged correctness, metadata) that leaderboards reduce to a single score and then discard. We reuse those artifacts to audit the
benchmark itself, and add an audit layer to VLMEvalKit: six detectors that run as post-processing over saved outputs, plus one additional text-only rerun with visual inputs removed, emitting a per-question report. The layer reports per-question findings together with aggregate audit statistics; for the headline quantities, it attaches permutation-null excess where applicable, interval estimates for proportions, and sensitivity to the model pool. Auditing MMStar, a
benchmark curated specifically so that the image is indispensable, with three models, we find zero aggregate accuracy gain from the image on 20.2% of its questions. An item-matched permutation null accounts for 17.0%, leaving a +3.2 pp excess (corrected permutation p = 0.001). This excess is concentrated in the mathematics and science categories and is absent from coarse perception, a localization consistent with knowledge substituting for visual evidence; confirming that mechanism would require item-level review. A second detector flags 134 questions on which every model that answered in parseable form picked the same non-gold option, and these candidates peak in the perception categories, whereas the zero-gain excess is largest in mathematics and science & technology. Replacing the LLM judge with exact string matching changes no conclusion, and every scoring discrepancy between the two traces to one model whose free-form answers a parser cannot read. On a binary yes/no
benchmark the detectors measurably fall out of calibration, a scope boundary that we measure and report. The audit layer, the prediction artifacts and the generated reports accompany this submission.
Read the full abstract ▾
All authors Denis Degtiarev, Anna Gromova, Sergey Katkov, Artem Mikhailin, Safina Mukhamedshina, Albert Nasybullin, Karina Shaikhutdinova, Artem Tuzov
Computer Vision
MOSAIC: Multi-Scale Object-Centric Slot Adaptation with Image-Guided Prototype Conditioning for Remote Sensing Image Super-Resolution
Liuqian Wang, Hanyu Xie, Rongzhen Zhao et al. (+6 more)
Pretrained
diffusion Transformers provide powerful generative priors for image restoration, yet adapting them to structure-sensitive tasks remains challenging. Existing methods mainly rely on dense low-resolution features as conditions, which often entangle salient objects with dominant backgrounds and provide limited abstraction for recurring structures across scales and samples. We propose MOSAIC, an object-centric representation learning framework that learns compact structural conditions for generative model adaptation. MOSAIC organizes degraded observations into structural slots that capture complementary object-level and layout-level evidence, integrates them across resolutions, and retrieves recurring priors from a shared prototype
memory to complement incomplete image-specific information. The resulting representations are injected into selected layers of a frozen
diffusion Transformer through lightweight residual modulation, preserving its pretrained generative prior. We instantiate MOSAIC for remote sensing image super-resolution, where small objects, repetitive structures, and large-scale variation provide a challenging testbed. Experiments on UCMerced, AID, and RSSCN7 show consistent PSNR improvements across $\times 2$ and $\times 4$ settings and favorable LPIPS results at $\times 4$. Qualitative results further demonstrate improved preservation of small objects, structural boundaries, and repetitive spatial patterns.
Read the full abstract ▾
All authors Liuqian Wang, Hanyu Xie, Rongzhen Zhao, Qi Ming, Chunjie Ma, Wentao Chao, Yufei Gao, Chengming Liu, Lei Shi
Emerging Technologies
Formalisation of information processing and intelligence in proteinoid networks
Andrew Adamatzky and Panagiotis Mougkogiannis
Proteinoid microspheres are electrically active assemblies of thermally polymerised amino acids that exhibit spontaneous oscillations, spike-like events, stimulus-dependent responses, adaptation and collective dynamics. We formulate these observations as an applied-mathematical problem and develop a unified dynamical framework for information processing in proteinoid networks. Individual microspheres are represented by nonlinear excitable or oscillatory systems, whereas interactions are described by weighted adaptive graphs and, in the spatial limit, reaction–
diffusion dynamics. The framework combines signal propagation, attenuation, fading
memory, adaptive coupling, synchronisation, information transfer, reservoir readout, robustness and energetic constraints. We define a task-relative proteinoid intelligence functional from normalised measures of stimulus information,
memory, adaptation, directed information transfer, task performance and robustness, penalised by energetic and structural costs. Analytical results establish a joint-response information bound, an upper bound on linear
memory for stable linear networks, and existence of an optimal admissible coupling
matrix under compact constraints. Dimensionless groups identify the relative time scales and coupling regimes governing the problem in nonlinear and adaptive network dynamics.
Read the full abstract ▾
Computer Vision
Efficient Confidence-Aware Densification for 3DGS
Manuel Hofer, Thomas Köhler, Felix Windisch et al. (+3 more)
3DGS-MCMC has replaced the gradient-driven density control of 3D Gaussian Splatting with opacity-based sampling and relocation, and is now the de facto standard for prior-free densification under a fixed primitive budget. Because opacity stays high where the scene is already well reconstructed, this sampling sends most new primitives to the foreground and leaves the background under-reconstructed. At low budgets, prior-based methods such as ConeGS and EDGS outperform it, at the cost of a pre-trained radiance field or dense image matches and a separate initialization stage. We argue for a comeback of error-driven densification within the fixed-budget MCMC framework. Our confidence-aware densification keeps the fixed budget and gradual growth of 3DGS-MCMC, but places primitives by the image error instead of the opacity. New primitives are drawn mostly where screen-space gradients are large. We favor primitives with a low learned confidence, which marks regions the reconstruction still explains poorly, and relocate the primitives that contribute least to the rendered images to these places. Averaged over 24 scenes, our method outperforms 3DGS-MCMC in PSNR, SSIM and LPIPS at every budget from 100k to 2M primitives. At 100k, without a prior or an initialization stage, it recovers 0.84dB of the 1.05dB PSNR by which ConeGS leads 3DGS-MCMC, and from 500k on, it has the highest PSNR and SSIM of all fixed-budget methods we compare with, including ConeGS and EDGS. While inherently prior-free, our densification can also exploit priors: from the ConeGS or EDGS initialization, it matches ConeGS in PSNR and SSIM at 100k and is best in all three metrics at 500k.
Read the full abstract ▾
All authors Manuel Hofer, Thomas Köhler, Felix Windisch, Bernhard Kerbl, Markus Steinberger, Lukas Radl
Numerical Analysis
On a maximum principle in discretization with spherical harmonics
Arnold Reusken
Spherical harmonics have been studied extensively and are used in many fields of application. We consider the probably earliest application of spherical harmonics, namely for a representation of the solution of the Laplace equation on the $d$-dimensional unit sphere where the boundary data are represented in a spherical harmonics basis. We consider general $d \geq 2$. For the solution $u$ of the Laplace equation the fundamental maximum principle holds. If one approximates the boundary data in the finite dimensional space spanned by all spherical harmonics of degree at most $N$, the resulting discrete solution, denoted by $u_N$, does in general not satisfy the maximum principle. In this paper we derive sharp estimates for $r_N^{\rm max}$ such that the maximum principle holds for $u_N(x)$ for all $x \in \Bbb{R}^d$ with $\|x\| \leq r_N^{\rm max}$. We show that for $d=3$ one has $r_N^{\rm max} \gtrsim 1- 3 N^{-1} \ln\big( 2^\frac13 N \big) $. Similar estimates are derived for $d \neq 3$. We also study the sharpness of these estimates.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from stat.ML
Defence against the dark arts of distillation: Unsupervised discovery of subliminal learning
Joseph Egan, Ciarán M. Gilligan-Lee, Michael O'Riordan et al. (+1 more)
When a
language model is distilled from a teacher that shares its base parameters, the student can inherit behavioural traits of the teacher—including misalignment—even from data that is semantically unrelated to the trait (Cloud et al., 2026). All existing defences require knowing the trait. We remove this requirement and prove that a subliminally transferred trait can be recovered from the parameter updates alone, without supervision, by exploiting a single asymmetry: the trait is expressed consistently across diverse
datasets generated by the teacher, whereas task-specific and shared capabilities vary. The resulting
estimator returns a weight-space direction that can be amplified to discover and interpret the trait or projected out to suppress it. Applied to the subliminal-learning
benchmarks of (Cloud et al., 2026), our method recovers the transmitted animal-preference traits without any trait labels and suppresses their expression with almost no impact to task performance.
Read the full abstract ▾
All authors Joseph Egan, Ciarán M. Gilligan-Lee, Michael O'Riordan, Yuchen Zhu
Computer Vision
Knowing Whether, Not Just Where: Anchor-Probing Distillation for Open-Set Remote Sensing Visual Grounding
Zhihao Chang, Lei Zhai, Shuyuan Yang et al. (+1 more)
Open-set visual grounding in remote sensing requires models to determine not only where a queried object is located, but also whether the image provides sufficient evidence for the query at all. Existing closed-set grounding models, however, tend to return plausible boxes even for absent objects, blurring localization confidence with genuine Visual Support. We introduce AnchorProbe-Distill (APD), a teacher-student framework that recasts open-set grounding as Visual Support learning. APD builds on the insight that supported queries should induce coherent spatial anchors, whereas unsupported or mismatched queries tend to produce unstable or misleading localization patterns. APD-T therefore uses the Original Query as a Reference Anchor to interpret query perturbations: Semantic-Preserving Queries retain the annotated target, Near-Category Ambiguous Queries are resolved according to whether they form a reliable alternative anchor or collapse onto the Reference Anchor, and Far-Category Counterfactual Queries provide rejection supervision. This reference-aware
reasoning is distilled into APD-S, enabling unified localization and rejection from a single image–query pair. Experiments show that APD enables the resulting model to reject absent targets while also improving its localization of present targets.
Read the full abstract ▾
All authors Zhihao Chang, Lei Zhai, Shuyuan Yang, Zhixi Feng
Artificial Intelligence
MAGMA-Bench: Evaluating Long-Horizon Embodied Agents under Dynamic Tasks and Execution Failures
Loan Bernat, Matthieu Grard, Ariane Herbulot et al. (+1 more)
Long-horizon embodied
agents must maintain task-relevant information, coordinate heterogeneous
robots, and adapt their decisions as execution unfolds. Existing
benchmarks investigate dynamic task execution and embodied multi-
agent coordination, but provide limited evaluation of their intersection. We introduce MAGMA-Bench, an architecture-agnostic
benchmark for persistent task
reasoning in dynamic multi-
robot environments. MAGMA-Bench evaluates
agents across increasing task horizons and controlled online perturbations, including execution failures, task interruptions, and mission updates, with scenarios requiring coordination across
robots with heterogeneous capabilities. It uses a structured skill-level interface to separate task-level
reasoning from perception and low-level control. We evaluate locally deployable open-weight planners, multi-
agent state-tracking strategies, and frontier models. Even the strongest evaluated models fail on a substantial fraction of clean episodes, and their success rates decrease further under online perturbations. Supervised adaptation greatly improves a small model, but performance remains substantially lower on longer and multi-
robot missions. We release the
benchmark, 261,646 task-level training examples, and an architecture-agnostic expert-trace generation pipeline.
Read the full abstract ▾
All authors Loan Bernat, Matthieu Grard, Ariane Herbulot, Florent Lamiraux
Social and Information Networks
Teams without teams: a trace-based method for measuring team-like organization
Thomas Baillie-Dawe, Cécile Bothorel and Nicolas Jullien
Digital platforms provide increasingly detailed traces of collaborative activity, yet these traces do not directly reveal whether contributions remain largely independent or whether more structured forms of collective work emerge through interaction. This paper addresses the methodological problem of measuring such organization when the collective being studied is not defined as a team in advance. We propose a trace-based method for measuring teamness, understood as the degree to which collaborative activity expresses properties associated with teamwork. The method translates established dimensions of teamwork into behavioral indicators that can be recovered from digital traces. These indicators can be examined separately and summarized through a composite measure of their joint expression. It does not require predefined team membership and is designed to be adapted across collaborative settings. We illustrate the method using revision histories from 52,437 articles in the English Wikipedia's Medicine WikiProject, comprising more than 10 million human-authored edits. Using article-months as observation units, the application shows that different dimensions of teamlike organization display distinct patterns and that their joint expression can vary within the same articles over time. The contribution is a general measurement framework for examining graded and dynamic forms of collective organization directly from behavioral traces.
Read the full abstract ▾
Computer Vision
S-Space: Exploring Spatial Workspace in Multimodal Language Models
Zhiyuan Shi, Xiang Li, Xingzhuo Guo et al. (+8 more)
Multimodal language models (MLLMs) exhibit promising spatial
reasoning capabilities, yet the internal mechanisms supporting them remain poorly understood. We uncover S-Space, a linear representation subspace in which object-
token representations encode continuous coordinates along horizontal, vertical, and distance axes. Although we identify these axes by differentiating spatial-answer scores with respect to object-
token activations, S-Space is intrinsic rather than induced by specific data or
prompts, remaining readable under non-spatial questions. We show that S-Space functions as an internal spatial workspace of MLLMs. Perception encodes object locations into it, forming spatial representations that can be directly read out to support accurate spatial judgments. Internal
reasoning transforms it for changing viewpoints. Verbal responses selectively access it, evidenced by causal interventions that alter spatial judgments while largely preserving non-spatial responses. Case studies further show that S-Space assigns locations to unseen entities and the observer, and connects physical directions with abstract linguistic meanings, suggesting both the capabilities and limitations of large-scale
language training. Finally, we use S-Space as a diagnostic tool for spatial
reasoning failures. Direction-word readouts reveal systematic coupling between cardinal and image-relative reference frames. More importantly, a perception-computation decomposition exposes a key bottleneck: MLLMs can form capable spatial representations, yet fall short in the spatial transformations required to reason reliably across viewpoint changes.
Read the full abstract ▾
All authors Zhiyuan Shi, Xiang Li, Xingzhuo Guo, Shangchen Miao, Weiliang Chen, Qiyu Dai, Jinshan Ren, Jiangran Lyu, Fangfu Liu, Mingsheng Long, Jialong Wu
Networking
On the Alignment Between Spectral Efficiency and Positioning Accuracy in Multi-LEO Clusters
Konstantinos Kapoglis, Sharief Saleh, Henk Wymeersch et al. (+1 more)
Low Earth Orbit (LEO) satellite constellations are expected to play a key role in future 6G Non-Terrestrial Networks (NTNs), where dense deployments enable cooperative multi-satellite processing for communication and positioning. While existing works have addressed communication and localization objectives separately, the correlation and
alignment between these objectives remain largely unexplored. In this paper, we investigate the correlation between Spectral Efficiency (SE) and positioning accuracy in cooperative multi-LEO satellite clusters. We consider an uplink direct-to-device scenario in which multiple satellites jointly process user signals and evaluate cluster performance using SE and Position Error Bound (PEB) metrics. We derive Time-of-Arrival (TOA)-based and Doppler-based PEBs, while communication performance is quantified through the achievable SE under joint signal processing. Using realistic LEO constellation simulations, we demonstrate that for clusters of seven satellites, SE exhibits a strong negative correlation with Doppler-based PEB (correlation coefficient of $-0.72$). In contrast, its correlation with TOA-based PEB is notably weaker, ranging between $-0.55$ and $-0.37$. These findings suggest that communication-oriented LEO clusters are largely compatible with Doppler-based positioning, but not necessarily with TOA-based positioning.
Read the full abstract ▾
All authors Konstantinos Kapoglis, Sharief Saleh, Henk Wymeersch, Hazem Sallouha
Machine Learning
Simulating Training Spikes of Deep Neural Networks by Layer-wise Random Feature Models
Ran Tian and Courtney Paquette
Intermittent “spikes” in loss, during the pre-training of Large
Language Models (LLMs), remains a mysterious phenomenon both theoretically and empirically. In this paper, we show that simple quadratic models with heavy-tailed, power-law distributed inputs can replicate these spikes. We introduce a theoretical framework and simulation
algorithm to model per-layer training dynamics of Deep Neural Networks (DNNs), using such quadratic models. In a case study on an Adam hyperparameter configuration (where $\beta_2$ is set close to $1$) known to induce spikes in
Transformer LMs, our simulation successfully predicts bumps indicating spiky behavior from a given checkpoint. Furthermore, we derive an appropriate $\beta_2$ schedule from observed metrics, which empirically improves performance.
Read the full abstract ▾
Robotics
Beyond Successful Robot Demonstrations: Learning from Human Interactions and Modeling Failure Dynamics
Peiyan Tu, Cong Wang, Hanxin Zhu et al. (+2 more)
Action-conditioned world models offer a promising foundation for generalist robotic intelligence by predicting the consequences of actions. However, existing robotic
datasets provide limited coverage of the diverse interaction dynamics required for generalizable world modeling. Large-scale egocentric human
videos offer complementary interaction experience, yet the embodiment gap between human hands and robotic manipulators hinders effective cross-embodiment learning. Meanwhile, the predominance of successful demonstrations leaves failure dynamics substantially underrepresented, limiting the ability of world models to anticipate unsuccessful executions. To address these limitations, we introduce HF-WM, an action-conditioned world model that combines Human interaction learning with Failure-aware post-training. To bridge the embodiment gap, we develop a dual-branch framework that couples
video generation with auxiliary pointflow prediction, using stabilized pointflow as a unified motion representation across human and
robot interactions while suppressing egocentric camera motion. To improve failure modeling, we further perform Group Relative Policy Optimization (GRPO) post-training on perturbed actions derived from successful demonstrations, guided by action
alignment and physical plausibility rewards, enabling the model to learn plausible failure dynamics without paired ground-truth
videos. Extensive experiments demonstrate that HF-WM achieves state-of-the-art performance in action-conditioned
video generation and exhibits strong utility for policy evaluation in downstream robotic tasks.
Read the full abstract ▾
All authors Peiyan Tu, Cong Wang, Hanxin Zhu, Xin Jin, Zhibo Chen
Artificial Intelligence
Beyond Domain-Routed Supervision: Cross-Teacher Guarding for Multi-Teacher On-Policy Distillation
Shi-Yu Tian, Yong Liu, Ximan Liu et al. (+10 more)
Multi-teacher on-policy
distillation (MOPD) extends on-policy
distillation (OPD) by routing domain-specific data to specialized teachers, integrating their capabilities through dense
token-level supervision on student-generated trajectories. However, domain routing only distinguishes the supervision sources for different domains; updates from all domains still act on shared parameters. When teachers differ substantially in response style and domain expertise, supervision that is effective in isolation may still lead to persistent gradient conflicts and capability degradation during joint optimization. To address this challenge, we propose Cross-Teacher Guarding for MOPD (Road MOPD), which introduces non-routed teachers as auxiliary cross-domain critics on the same student trajectories and coordinates multi-teacher supervision through
token-level credit assignment and gradient updates. Specifically, cross-teacher-guided credit assignment uses teacher agreement to strengthen
token-level corrections jointly supported by the teachers and attenuate conflicting signals; preventive multi-teacher gradient projection uses auxiliary gradients from the same trajectories to constrain potentially conflicting updates and adaptively controls projection strength to retain target-domain learning signals. In experiments, Road MOPD mitigates degradation in the disadvantaged domain, gains over 5 percentage points under high conflict, and improves overall performance in low-conflict and three-domain settings.
Read the full abstract ▾
All authors Shi-Yu Tian, Yong Liu, Ximan Liu, Tianrui Zhu, Sule Bai, Xinwei Yang, Ruiqi Liu, Kun-Yang Yu, Ming Yang, Zhi Zhou, Yu-Feng Li, Yipeng Sun, Gu Simiu
Artificial Intelligence
Causal Effect Identifiability in Causal Abstractions: A Strict Hierarchy of Criteria
Clément Yvernes, Marianne Clausel, Emilie Devijver et al. (+1 more)
Identifying causal effects from observational data typically requires a fully specified causal graph, an assumption that is often unrealistic in complex or high-dimensional settings. Causal abstractions address this limitation by retaining only partial causal information, thereby defining equivalence classes of compatible causal graphs. In this paper, we study causal effect identifiability directly at the level of these graph collections, rather than at the level of individual causal diagrams. We formalize several notions of identifiability in this setting and organize them into a hierarchy that clarifies their relationships and differences. In particular, we establish that a causal effect may be identified over a collection of graphs without admitting a single causal derivation that holds uniformly across all compatible graphs. These results help reveal gaps and clarify ambiguities in the existing literature.
Read the full abstract ▾
All authors Clément Yvernes, Marianne Clausel, Emilie Devijver, Eric Gaussier
Machine Learning · Cross-listed from math.DS
Asymptotic Analysis of Discrete-Time Self-Attention Dynamics in Transformers
Claudio Altafini
In machine learning, a self-
attention dynamics describes the
attention mechanism of
transformers, and is obtained treating the layer index of a
transformer as time index. Even though naturally discrete-time, this dynamical system is most often studied in continuous-time. Scope of this paper is to provide a thorough investigation of the discrete-time version of the self-
attention dynamics: a complete classification of its
equilibria is given, and their stability properties are studied, in both the single-head and the multihead case. We show that in both cases the main attractor of the dynamics is a consensus
equilibrium aligned with the principal eigenvector of the value
matrix (in the single-head case, a corresponding
matrix of weights in the multihead case), even though other locally asymptotically stable bipartite
equilibria can be show to exist at least in the single-head situation. The behavior is confirmed by looking at the learned weight
matrices of a real
transformer.
Read the full abstract ▾
Robotics
RESCUE: Is This Alarm Worth Acting On? Counterfactual Evaluation of VLA Runtime Monitors
David Ahmedt-Aristizabal, Abdelwahed Khamis, Mohammad Ali Armin et al. (+1 more)
Runtime failure monitors for
robot policies are typically evaluated by how accurately they distinguish failed from successful executions. Yet an accurate alarm may arrive too late for recovery, while an unnecessary intervention may disrupt an execution that would otherwise succeed. We introduce RESCUE, a counterfactual framework for evaluating whether a monitor's alarms enable useful interventions. By restoring simulator and policy states at execution checkpoints, RESCUE compares alternative actions, identifies the latest checkpoint satisfying a specified recovery criterion, and evaluates whether acting on alarms reduces expected task-completion cost. We evaluate RESCUE with OpenVLA-OFT and $\pi_{0.5}$ over LIBERO-Spatial and LIBERO-Plus using 128,340 counterfactual branches. Recoverability depends on the state–intervention pair rather than on
robot state alone: at $\tau=0.5$, 14 of 66 OpenVLA-OFT/LIBERO-Plus failures contain a recoverable checkpoint under continuation, compared with 34 of 66 under continuation or scripted recovery, while recovery reduces recoverability in other conditions. On the primary strong-policy conditions, favourable detection scores do not establish lower intervention cost. Under the evaluated stop-and-reset model, even an idealised detector can afford only a 2.0–3.3% false-alarm rate before unnecessary resets outweigh the savings from stopping failures early. These results show that monitor value depends jointly on the intervention available and the workload in which it is deployed.
Read the full abstract ▾
All authors David Ahmedt-Aristizabal, Abdelwahed Khamis, Mohammad Ali Armin, Lars Petersson
Artificial Intelligence
Retrieval-Augmented Multimodal Language Model with Heterogeneous Evidence Integration for Clinical Prognosis
Hamza Shafiq, Jialu Tang, Ting Dang et al. (+2 more)
Clinical prognosis requires integrating structured measurements, physiological waveforms, longitudinal history, and population context. We introduce PRISM (Population-Retrieved Inference for Structured
Multimodal Prognosis), a
multimodal language model that combines per-feature numerical tokenization, recency-weighted aggregation of longitudinal ECGs, and a dual-modality population
memory bank.
Retrieval operates in representation space: the model receives mean-pooled
embeddings of similar training-cohort encounters, while no individual retrieved record enters the
prompt. On the 1,443-task MDS-ED
benchmark, PRISM achieves AUROCs of 86.2% for diagnosis, 93.1% for deterioration, 91.8% for ICU admission, and 92.8% for mortality. These results use a 96-hour pre-arrival window unavailable to reported baselines. With
benchmark-window inputs only, PRISM reaches 0.8895 AUROC, compared with 0.8937 for UniPACT and 0.8889 for MDS-ED. A matched-input compact fusion model using the same five evidence streams and frozen encoders reaches 0.7652 AUROC, versus 0.9092 for PRISM. Replacing numerical EHR encoding with text serialization reduces AUROC by 3.66 points; removing pre-arrival history reduces it by 1.97 points; and zeroing
retrieval reduces it by 0.99 points. Genuine
retrieval outperforms random-neighbor
retrieval by 0.97 macro-average points, although a paired task-level sign test does not establish a uniform advantage (p=0.067). On the external MC-MED cohort, supervised fine-tuning reaches 0.8611 AUROC, compared with 0.8227 for locally trained gradient boosting and 0.7841 for source-only PRISM. These results support structure-aware
multimodal integration and supervised external adaptation.
Read the full abstract ▾
All authors Hamza Shafiq, Jialu Tang, Ting Dang, Jun Hu, Aaqib Saeed
Machine Learning
FedPCD: Priority-Constrained Federated Aggregation
Yousef Irshaid, Dara Varam and Mohamed I. AlHajri
Federated learning (FL) seeks to optimize a shared global model from decentralized client data. While effective for average performance, this process can produce unfair performance across clients, particularly under statistically heterogeneous data distributions. Existing
fairness-aware and multi-objective FL methods address this through client re-weighting, gradient correction, or common-descent formulations, but often change the global optimization direction or treat client objectives symmetrically. We propose FedPCD, a priority-constrained
federated optimization framework that separates the global optimization objective from client protection. FedPCD preserves the conventional
federated aggregate as the primary objective, while allowing only the minimum correction necessary to meet client performance constraints. Across heterogeneous FL
benchmarks, FedPCD consistently outperforms state-of-the-art baselines, achieving the highest global accuracy across all configurations and leading worst-10% client accuracy in nearly all. FedPCD further preserves competitive trade-offs between global accuracy and client performance variability. Beyond this trade-off, FedPCD demonstrates faster empirical convergence, reaching matching or superior performance in substantially fewer communication rounds.
Read the full abstract ▾
Sound
Lossless Inference Acceleration for ASR via EAGLE-Style Speculative Decoding
Sen Zhang, Xianghu Yue, Yiming Chen et al. (+3 more)
Autoregressive
Transformer and Conformer models have achieved state-of-the-art performance in automatic
speech recognition (ASR). However, their sequential
token generation mechanism imposes a significant bottleneck during decoding, resulting in substantial inference latency. While speculative decoding has emerged as a promising lossless acceleration technique, existing ASR-oriented approaches either suffer from low draft acceptance rates due to suboptimal draft quality or rely on specially designed external draft models, which limit their generality and practicality. To overcome these limitations, we propose EAGLE-ASR, a novel framework that accelerates ASR inference via EAGLE-style speculative decoding while strictly preserving lossless output quality. Specifically, we introduce
Speech-Draft, a lightweight autoregressive module that serves as a universal draft model by jointly leveraging encoder acoustic features and hierarchical decoder representations from the target ASR model. Furthermore, we adopt a multi-step generation strategy with self-distilled data to train the
Speech-Draft, enabling it to better approximate the target model's output distribution and maintain causal consistency during multi-
token generation. Extensive experiments on representative Whisper and Conformer AED models demonstrate that EAGLE-ASR achieves significant lossless speedups of up to $1.9\times$ and $2.4\times$, respectively, consistently outperforming existing speculative decoding baselines for ASR. Our proposed framework further generalizes effectively to multilingual ASR and
speech translation tasks, establishing EAGLE-ASR as a practical and universally applicable solution for accelerating modern ASR systems without compromising recognition accuracy.
Read the full abstract ▾
All authors Sen Zhang, Xianghu Yue, Yiming Chen, Wenhuan Lu, Jianguo Wei, Haizhou Li
Machine Learning
Beyond Temporal Proximity: Explanatory World Models for Credit Assignment in Reinforcement Learning
Oriol Corcoll, Qiyao Wei and Ciarán Mark Gilligan-Lee
Generative Recommenders increasingly use Reinforcement Learning (RL) to optimize outcomes that may appear hundreds or thousands of interactions after a decision. Standard RL propagates reward according to temporal proximity rather than explicit relevance, attenuating the signal assigned to distant but important actions and creating a source of sample inefficiency. We introduce Explanatory World Models (EWM), an attribution framework that uses explanations to identify the earlier events that plausibly contributed to a realized outcome. We instantiate EWM with a frontier
language model and use its explanations as learning signals using a conservative hierarchical operator that recursively redistributes reward while preserving the undiscounted episode return. To investigate what constitutes a useful explanation, we operationalize three complementary variants inspired by Halpern's actual causation, Lombrozo's explanatory simplicity, and Deutsch's hard-to-vary explanations. We evaluate EWM on two controlled long-horizon RL
benchmarks, a public sequential-recommendation
benchmark, and an industry-scale generative recommender. On Crafter and WordCraft, EWM substantially outperforms PPO and LaRe, with its largest gains when rewards are most delayed. On KuaiRand, EWM outperforms SFT and LaRe offline and A2C online. Finally, in an offline evaluation of an industry generative recommender, EWM considerably improves Recall and increases behavioral proxies. These results suggest that explanations can provide structured and reusable credit for learning from delayed outcomes.
Read the full abstract ▾
Computer Vision
SplatQ: Interaction-Aware Pruning of 3D Gaussian Splatting via Quadratic Selection
Hongbin Ye, Haixu Liu, Shenwei Kang et al. (+13 more)
Existing 3D Gaussian Splatting (3DGS)
pruning methods primarily rely on global Top-K selection based on independent importance scores, lacking regional budget control and ignoring joint deletion effects. Consequently, they cannot ensure balanced reconstruction quality across a scene. We propose SplatQ, a combinatorial
pruning algorithm that coordinates regional budget allocation and joint Gaussian selection under a fixed budget. SplatQ partitions Gaussians by spatial proximity and appearance similarity into size-limited blocks, decomposing global selection into subproblems of a specific size. It then smooths importance-based block retention quotas to prevent excessive budget concentration. Finally, it constructs a quadratic approximation of joint deletion effects using the responses of representative pixel colors to changes in Gaussian opacity. Individual responses, pairwise interactions, and quota penalties are incorporated into a quadratic unconstrained binary optimization (QUBO) objective, solved using simulated annealing (SA) or a coherent Ising machine (CIM) to jointly select Gaussians. Experiments on 5
benchmarks demonstrate that SplatQ outperforms state-of-the-art
pruning methods in aggregate reconstruction quality, spatial quality balance, and rendering speed. Ablations confirm that budget smoothing improves local detail reconstruction, while interaction modeling improves Gaussian selection quality under a fixed budget.
Read the full abstract ▾
All authors Hongbin Ye, Haixu Liu, Shenwei Kang, Wenxiu Wei, Tinglan Wang, Hongdong Zhu, Fan Tang, Qi Gao, Yin Ma, Kai Wen, Hai Wei, Nuo Qian, Wei-Shi Zheng, Shaobo Chen, Fa-Ting Hong, Jingke Meng
Computer Vision
Stride: Scale-Aware Multi-Agent Evolutionary Refinement for Infographic Generation
Yan Zhou and Yang Yang
Infographic generation challenges text-to-image models because a single image must satisfy numerous visual, textual, and semantic constraints. Iterative agentic refinement with a
multimodal verifier is a natural remedy, but existing methods face a scale mismatch: local edits cannot correct global composition errors, while full regeneration may destroy already-correct regions. We introduce Stride, a constraint-grounded multi-
agent framework that treats refinement over any frozen generator as a history-aware
evolution strategy. Stride separates constraint evaluation - whether each requirement is satisfied - from visual grounding - where and how it is violated. It then routes global revisions to a composer subagent and local corrections to a retoucher subagent. Each subagent conducts internal planning and generates multiple candidates, while Stride accepts a candidate only when it improves the verifier score, reducing regression across iterations. Because
stochastic generators and imperfect verifiers make any single step unreliable, it is this combination of scale-aware routing, per-step redundancy, and monotonic acceptance that yields stable improvement across iterations. With fully open-weight generator-verifier systems, Stride outperforms proprietary systems including Nano-Banana-2.0 and Seedream-5.0-Pro on BizGenEval, matches GPT-Image-2.0 and surpasses Seedream-5.0-Pro on IGenBench, and achieves a new state-of-the-art score on WISE, while substantially outperforming existing agentic methods across all evaluated
benchmarks.
Read the full abstract ▾
Sound
Phonology-Aware Speaker Extraction Frontend for Audio LLMs
Yanxiang Chen, Haoyuan Deng, Jinyu Wang et al. (+4 more)
Target speaker extraction (TSE) is a key frontend for enabling audio large
language models (LLMs) to recognize and understand a designated speaker in overlapping
speech. However, multilingual mixtures exhibit heterogeneous extraction demands, causing fixed-depth systems to either under-process challenging inputs or waste computation on inputs that have already converged. Existing TSE frontends generally apply uniform computation, while Audio-LLM adaptation methods neither explicitly model
language-conditioned extraction demands nor preserve plug-and-play compatibility with frozen downstream models. We therefore propose LATTE, a lightweight 5.2M-parameter frontend that combines an acoustic Difficulty
Estimator, a task-adaptive phonology-informed prior learned from multi-depth TSE gains, and an Adaptive Hourglass Extractor to route each mixture without requiring
language identity at inference. On PolyExtract, LATTE achieves a macro SI-SNRi of 15.07 dB at 0.30 GFLOPs, retaining performance within 0.09 dB of a 0.53-GFLOP fixed-depth model while reducing computation by 43.4%, and lowers the average WER from 16.8% to 14.6% relative to TF-GridNet. By jointly modeling sample-level extraction demands and phonological structure, LATTE provides a practical framework for quality-aware, compute-efficient multilingual TSE and offers a scalable path toward robust Audio-LLM deployment.
Read the full abstract ▾
All authors Yanxiang Chen, Haoyuan Deng, Jinyu Wang, Hongyi Chen, Dingshan Li, Runlin Huang, Weifeng Su
Hardware Architecture
A Hybrid and Hierarchical Ray Tracing Method for Efficient 3D Gaussian Rendering
Tiancheng Zhao, Sixu Li, Yuzhou Chen et al. (+3 more)
3D Gaussian Splatting (3DGS) is a powerful representation for real-time neural reconstruction and rendering. While existing 3DGS systems rely on rasterization for efficiency, recent work explores ray tracing to enable secondary lighting, highly distorted cameras, and depth-of-field effects. However, compared with rasterization, ray-traced 3DGS introduces three major challenges: (1) the highly non-uniform spatial distribution of massive numbers of Gaussian primitives creates severe load imbalance; (2) neighboring pixels often generate rays with highly similar traversal paths, leading to substantial redundant traversal overhead; and (3) supporting both Gaussian and triangle primitives yields heterogeneous intersection workloads that are not well matched to existing GPU architectures. We present H2-Ray, a hybrid and hierarchical ray tracing architecture for efficient 3D Gaussian rendering. H2-Ray introduces three techniques. Load-Balanced Spatial BVH (LB-SBVH) partitions the scene into uniform grids and uses density-aware representations to convert an imbalanced global workload into balanced local traversal units. Ray-Cluster Path Reuse (RCPR) lets a proxy ray share traversal state with neighboring rays to amortize redundant traversal. The Unified Intersection Engine (UIE) supports mixed Gaussian and triangle intersections through a mode-multiplexed datapath that shares 40% of its floating-point units across both primitive types. Across 8 scenes, H2-Ray achieves a
geometric-mean speedup of 3.77x and a maximum speedup of 9.26x over RaySplats on the NVIDIA RTX A5000, combining a cycle-accurate model of the modified RT Core with measured CUDA timings.
Read the full abstract ▾
All authors Tiancheng Zhao, Sixu Li, Yuzhou Chen, Hengyu Ding, Zhiming Pan, Chaojian Li
Systems and Control
Hierarchical Graph Neural Network for Probabilistic Power Flow with Grid Topology Contingencies
Jinbao Wang, Jun Liu, Shiliang Zhang et al. (+4 more)
Data-driven surrogates reduce the nonlinear solves required for probabilistic power flow (PPF), but models tied to a fixed bus–branch graph often generalize poorly when outages change electrical coupling. This paper proposes a
topology-decomposition-driven hierarchical bidirectional graph neural network (TDD–HBGNN) for PPF under unseen
topologies. For each realized
topology, the network is decomposed into fundamental topological units (FTUs), dynamically homogeneous voltage clusters (DHVCs), and a core backbone sub-grid (CBSG), with equivalent-admittance representations preserving electrical ports between units. A sensitivity-based linear prior captures the first-order voltage-magnitude and phase-angle response, while the HBGNN learns
topology-dependent nonlinear residuals. Impedance- and uncertainty-aware bottom-up aggregation condenses local perturbations, while mode-modulated top-down broadcasting restores system context at buses. We evaluate our approach on the IEEE 118-bus and Polish 2746-bus systems under unseen N–1 and N–2 contingencies against six baseline methods. Compared with these methods, TDD–HBGNN achieves 2.47–4.20-fold lower all-bus phase-angle errors and lower phase-angle errors in every held-out event. Its voltage-magnitude advantage is concentrated near the outaged branches, where paired error ratios reach 1.75–2.58 on the IEEE system and 1.87–1.97 in Polish N–2 events. All system–contingency–baseline comparisons favor TDD–HBGNN in joint voltage–angle error. Moreover, shared hierarchical operators keep the model size at 0.658 million parameters across both grids and yield 0.161–1.793~ms forward latency per probabilistic case. These results demonstrate system-wide angular accuracy, outage-local voltage accuracy, and parameter-efficient scaling across unseen
topologies.
Read the full abstract ▾
All authors Jinbao Wang, Jun Liu, Shiliang Zhang, Xuehui Ma, Lin Yan, Haolin Liu, Sabita Maharjan
Artificial Intelligence
Humanize MLE Agents: Multi-Agent Session Alternation Beyond Single-Agent Limits
Changye Li, Zijian Zhang, Muheng Li et al. (+12 more)
Existing LLM-based
agent approaches for long-horizon machine learning engineering (MLE) often use context engineering, implemented through custom harnesses that manage tools, skills, and accumulated experience. However, we find that as single-session execution lengthens, the
agent's
reasoning intensity declines alongside slowing progress, summarized as long-horizon deceleration, and the benefits of these custom harnesses vanish as the underlying model improves, summarized as harness staleness. These findings suggest that their benefits may not persist across versions, which motivates a workflow-level approach in which
agents build on one another's work across sessions. We model these alternating
agent sessions as a context-reset semi-Markov decision process (CR-SMDP) and establish sufficient conditions for alternation to exceed single-
agent reference limits, derive an explicit experiment-budget bound, and characterize when session caps preserve this advantage. Guided by this analysis, we introduce Humanize MLE
Agents (HMA), which alternates two
agents on a shared workspace, limits each session to at most k experiments and resets the context between sessions, without additional task-specific harness or prior knowledge supplements. Across 75 Kaggle tasks in MLE-bench, HMA reaches a state-of-the-art any-medal rate of 78.2%, exceeding the strongest open-source baseline by 8.0 points with a quarter of its time budget. Moreover, among six configurations spanning five heterogeneous
agent pairs, HMA improves any-medal rate over the constituent baselines by 6.4 points on average, demonstrating cross-pair generality. A supplementary case on kernel optimization reaches 878 cycles, within 1.6% of the community leaderboard leader, demonstrating applicability beyond MLE.
Read the full abstract ▾
All authors Changye Li, Zijian Zhang, Muheng Li, Yiduo Pang, Tingjun Wu, Menghan Li, Yitong Liu, Hongkai Guo, Mengchi Chen, Chenyu He, Kaisheng Ma, Wei Xu, Song Han, Sihao Liu, Ligeng Zhu
Cryptography and Security
EviRepair: Evidence-Driven Agents for Reliable Vulnerability Repair
Yilian Shi, Qingyu Liu, Tianyi Zhou et al. (+5 more)
Recent advances in large
language models (LLMs) have enabled
agent-based approaches for repository-level automated vulnerability repair (AVR), showing promising repair capabilities. However, an important issue has received limited
attention: how can we ensure that the evidence generated during the repair process is reliable enough to support accurate vulnerability understanding and patch generation? In this paper, we propose EviRepair, an evidence-driven multi-
agent framework for repository-level AVR. EviRepair introduces evidence review into the repair process by enabling
agents to collaboratively collect vulnerability evidence, verify evidence consistency, and construct a reviewed vulnerability model for patch synthesis and validation. Extensive experiments on PatchEval and SecBench
benchmarks with multiple
agent frameworks and LLMs demonstrate that EviRepair consistently outperforms existing
agent-based baselines, achieving average improvements of 16.37% and 40.33% in resolved rates on PatchEval and SecBench, respectively. Codes are available at https://github.com/lly-s/EviRepair
Read the full abstract ▾
All authors Yilian Shi, Qingyu Liu, Tianyi Zhou, Wende Tan, Chenyang Si, Yueming Lyu, Xia Hu, Dongrui Liu
Computation and Language
Proactive Agentic Orchestration
Xueyang Feng, Chaozhuo Li, Xiaohe Bo et al. (+4 more)
Agentic orchestration delegates tasks to sub-
agents with independent contexts, providing a key path to extending the effective execution horizon of large
language model
agents. However, existing orchestration methods face two limitations: in execution, current methods fix sub-
agents' goals and harnesses at creation time, limiting their adaptability to dynamic real-world environments; in optimization, sparse and delayed feedback constrains effective learning of the orchestration policy. To address these limitations, we introduce POrchestra, an on-the-fly proactive agentic orchestration framework. It allows sub-
agents to proactively communicate with the orchestrator at any logical point to request harness adjustments while maintaining continuity of local execution. This mechanism not only enables sub-
agents to adapt seamlessly to dynamic environments but also naturally inspires On-Policy Orchestration
Distillation (OPOD) based on hindsight feedback. OPOD attributes sub-
agents' proactive revision requests back to preceding orchestration decisions to derive targeted improvement signals, and leverages these signals as process supervision alongside the final reward to optimize the orchestration policy. We evaluate POrchestra on three long-horizon
agent benchmarks spanning deep research, dynamic digital assistance, and software engineering. POrchestra improves task success by an average of 9.1% relative to the strongest baseline; combined with OPOD, it improves small-orchestrator success by an average of 10.0% relative to existing training baselines. POrchestra also achieves a more favorable cost-performance trade-off. Our code is publicly available at https://github.com/XueyangFeng/POrchestra.
Read the full abstract ▾
All authors Xueyang Feng, Chaozhuo Li, Xiaohe Bo, Jiakai Tang, Chenxi Li, Qiwei Ye, Xu Chen
Numerical Analysis
Regularisation and Parameter Choice Rules for Composite Linear Inverse Problems
Christina Brandt and Patrick Horn
In this paper we propose a novel regularisation strategy for structured linear inverse problems of the form $Ax=F \circ G x = z$. Instead of using the operator $A$ or a complex scheme involving both $F$ and $G$ simultaneously, combining already existing and simpler regularisations for $F$ and $G$ allows the use of established and optimised programs for high-speed computations. As classical parameter choice rules fail in the scenario when no source conditions are known because intermediate error bounds may be unavailable, we provide both a-priori and a-posteriori rules with a convergence speed analysis given source conditions. We test the proposed method on the pre-clinical medical imaging technique of
magnetic particle imaging with different choices of regularisation and on a convolution problem for an analysis of the convergence speed.
Read the full abstract ▾