Artificial Intelligence
Mini-Batch Risk-Averse Deep Q-Learning: A Robot Navigation Case Study
Aayush Patel and Andrzej Ruszczyński
We study the control of Markov decision processes in which the quality of a policy is evaluated by a dynamic, time-consistent Markov risk measure rather than by an expected discounted cost. The main obstacle to combining such measures with reinforcement learning is that a transition risk mapping depends on the transition kernel in a nonlinear way, and therefore cannot be estimated from a single observed transition. We remove this obstacle by employing mini-batch transition risk mappings: the mapping is applied to the empirical measure of $N$ independent next-state samples, and the result is averaged. The resulting mapping is again coherent. However, as an expected value of a function of $N$ next-state values, it admits an unbiased one-sample
estimator. We embed this mapping into a double deep Q-network, analyze the two sources of estimation bias that arise, and obtain a risk-averse Q-learning method applicable to state spaces far beyond the reach of tabular schemes. The method is applied to an
underwater robot navigation problem, in which a vehicle must visit collection points, gather
stochastic information payloads, and deliver them at transmission points, while exposed at each step to the risk of destruction. A hierarchical decomposition delegates path execution to an exact graph search and confines learning to the high-level “collect or transmit” decision. A low-dimensional feature map, invariant under the
symmetries of the problem, replaces the raw state–configuration encoding. In experiments on $300$ held-out environments, the resulting policies transfer to instance sizes never seen in training, and already $N=2$ reduces the upper semideviation of the outcome distribution while simultaneously improving its mean whenever the simulator is misspecified—an empirical counterpart of the duality between coherent risk measures and distributional robustness.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.LG
HyCO: A Hybrid Neural Solver for Combinatorial Optimization
Yuheng Li, Di Yang, Haipeng Chen et al. (+1 more)
Sequential reinforcement learning (RL) solvers and global
diffusion model (DM) solvers for neural combinatorial optimization exhibit complementary failure modes under an optimization-regret view. The former enjoys small marginal regret in the early construction stage, but suffers from horizon-wise compounding errors with super-linear regret growth; the latter avoids horizon compounding but incurs linear or sublinear regret w.r.t. the dimension of the remaining unsolved subspace. We propose Hybrid Neural Solver for Combinatorial Optimization (HyCO), a hybrid inference
algorithm that constructs a solution prefix with an RL solver and adaptively switches to a conditional DM to complete the remaining decisions. To characterize why such hybridization helps, when to trigger the handover, and how to realize it in practice, we first develop a unified error-scaling theoretical framework and prove that, under explicit error-scaling assumptions, i) the hybrid structure achieves strictly lower expected regret than either backbone alone, and ii) there exists a unique optimal trigger step that minimizes the hybrid regret. We then design a lightweight adaptive trigger that combines policy
entropy and RL-DM disagreement to detect trajectory-level signals of the regime shift as a practical proxy, since the optimal trigger step is defined at the expected-regret level and is not directly computable on individual trajectories. Experimental results on diverse
benchmarks demonstrate that HyCO achieves consistent improvements over both backbones and support the empirical effectiveness of adaptive triggering.
Read the full abstract ▾
All authors Yuheng Li, Di Yang, Haipeng Chen, Yanhai Xiong
Artificial Intelligence
When Can LLM Digital Twins Reduce Human Measurement? From Behavioral Fidelity to Statistical Substitutability
Steven Wang, Kyle Hunt, Shaojie Tang et al. (+1 more)
LLM-based digital twins promise to reduce repeated human data collection by generating person- specific responses, yet existing evaluations provide little evidence about whether they can reduce human measurement while preserving valid inference. To address this, we introduce statistical substitutability, an inferential criterion that evaluates the extent to which twin predictions can reduce human measurement for a particular estimand while preserving valid inference. We develop a framework, grounded in mixed-subject and prediction-powered inference, that evaluates statistical substitutability along four dimensions: aggregate fidelity, paired respondent-level signal, finite-sample human-label recovery, and stability across populations. Across two empirical evaluations spanning behavioral experiments, multiple models, and alternative respondent representations, we find that digital twins can reproduce average human effects while providing little information about which individuals differ from those averages. Newer models and richer respondent information improve some dimensions of performance but do not reliably translate into human-data savings. Human calibration can reduce aggregate prediction error, yet limited labeled samples often fail to produce stable precision gains. Importantly, these findings demonstrate that behavioral fidelity is neither necessary nor sufficient for statistical substitutability. More broadly, they suggest that AI-generated evidence should be evaluated based on its ability to support valid scientific inference rather than its ability to reproduce human outcomes alone. Digital twins should therefore be judged for confirmatory use by whether they reduce uncertainty about human quantities, not merely by whether they reproduce human means, distributions, or effects.
Read the full abstract ▾
All authors Steven Wang, Kyle Hunt, Shaojie Tang, Kenneth Joseph
Artificial Intelligence
From Event Logs to Governed Action: A BlueSky Agenda for Agentic Process Mining
Yiyuan Yang, Zheshun Wu, Yong Chu et al. (+3 more)
Process mining has long turned event logs into process knowledge: discovered models, conformance evidence, bottleneck diagnoses, and runtime predictions. Agentic AI changes the target. Process-aware
agents will not only ask what happened. They will ask whether a proposed action should be taken, given the available evidence,
privacy budget, organizational authority, and downstream risk. This BlueSky paper proposes event-to-action process mining: a process-mining agenda for transforming heterogeneous operational event data into governed action. The goal is not another dashboard, a generic enterprise simulator, or a
language interface over logs. We argue that the community needs four mineable artifacts: event-object representations, action evidence packages, governance contracts, and
benchmarks where act, defer, ask, and refuse are all valid outputs. This agenda is timely because agentic business process management (BPM), LLM-assisted process mining, object-centric event standards, causal process monitoring, and
privacy-preserving learning are maturing separately. Bringing them together defines a data-mining target inside process mining: mining logged organizational behavior for accountable action, not only retrospective insight.
Read the full abstract ▾
All authors Yiyuan Yang, Zheshun Wu, Yong Chu, Zhenghua Chen, Zenglin Xu, Qingsong Wen
Artificial Intelligence · Cross-listed from cs.CL
Rethinking Sign Language Translation: The Impact of Signer Dependence on Model Evaluation
Keren Artiaga, Sabyasachi Kamila, Haithem Afli et al. (+2 more)
Sign
Language Translation has advanced with deep learning, yet evaluations remain largely signer-dependent, with overlapping signers across train/dev/test. This raises concerns about whether models truly generalise or instead rely on signer-specific regularities. We conduct signer-fold cross-validation on GFSLT-VLP, GASLT, and SignCL, three leading, publicly available, gloss-free SLT models, on CSL-Daily and PHOENIX14T. Under signer-independent evaluation, performance drops sharply: on PHOENIX14T, GFSLT-VLP falls from BLEU-4 21.44 to 3.59 and ROUGE-L 42.49 to 11.89; GASLT from 15.74 to 8.26; and SignCL from 22.74 to 3.66. We also observe that in CSL-Daily many target sentences are performed by multiple signers, so common splits can place identical sentences in both training and test, inflating absolute scores by rewarding recall of recurring sentences rather than genuine generalisation. These findings indicate that signer-dependent evaluation can substantially overestimate SLT capability. We recommend: (1) adopting signer-independent protocols to ensure generalisation to unseen signers; (2) restructuring
datasets to include explicit signer-independent, sentence-disjoint splits for consistent benchmarking; and (3) reporting both signer-dependent and signer-independent results together with train-test sentence overlap to improve transparency and comparability.
Read the full abstract ▾
All authors Keren Artiaga, Sabyasachi Kamila, Haithem Afli, Conor Lynch, Mohammed Hasanuzzaman
Artificial Intelligence
Support Topology and Gradient Mixing in Sinkhorn Layers
Dylan Forde
Sparse Sinkhorn layers use a fixed support graph to restrict transport between
tokens. How does this graph control gradient propagation through the scaling iterations. We develop a fixed-support calculus showing that each row-column cycle induces a row-
stochastic operator on column-potential perturbations modulo constants. Its transpose propagates zero-mass reverse-mode cotangents. The finite-cycle operator uses two distinct half-step transport plans; at a balanced fixed point it reduces to a two-step walk determined by a single plan. We derive the accompanying score and marginal source terms and use Dobrushin contraction and minorization to bound homogeneous and source-driven tail cotangents. Our main result characterizes when support and marginals guarantee one-step contraction uniformly over finite scores: every feasible face of the transportation polytope must have pairwise two-hop column overlap. Otherwise, suitable score directions make the contraction coefficient arbitrarily close to one. We extend this analysis to ordered support schedules and derive certificates for partition heat-bath layers, coordinate sweeps, forced shared mass, and register-augmented supports. These results provide mathematical criteria for support design in differentiable transport layers, with guarantees restricted to the fixed-support quotient-gradient component.
Read the full abstract ▾
Artificial Intelligence
CausalVerify: End-to-End Verification of Causal Analyses by Language Models
Yonghong Zhang, Ricardo Correia, Isabel M. Parra et al. (+1 more)
Language models increasingly perform empirical analyses end to end, yet existing evaluations assess the written explanation or whether generated code executes, not whether the executed workflow recovers the intended causal estimand. We introduce CausalVerify, an execution-grounded
benchmark for end-to-end causal analysis that follows a model from research-context interpretation to estimand recovery. It scores this workflow at four distinct layers: method recognition, design specification, executable implementation, and estimand recovery. It combines 259 real-paper contexts, 100 fixed-seed synthetic scenarios with executable reference estimates, and 23 paper-twin pairs in which a model commits to a design before seeing the data and its executed analysis is scored against a canonical
estimator on the same realised
dataset. Execution is not correctness. Among 426 model-written workflows that run without error, 15.5% fail verification, and a keyword score of the effect direction stated in the text is a poor proxy for recovery. In a 23-pair, nine-model paired study, replacing a model's committed design with the reference design and its execution conventions raises joint recovery of the point estimate and standard error from 15.0% to 51.5%; yet 48.5% of eligible seeds still fail under the reference design. The direction replicates on six pairs built afterwards under a frozen construction protocol, although on the two newest pairs the gain is confined to models from the family that built the references. Design specification is consequential but not sufficient: a plausible method and runnable code do not guarantee recovery, and even supplying the reference design and its conventions leaves substantial downstream failure. CausalVerify evaluates the executed workflow rather than its surface plausibility, and every reported number is recomputed from frozen artifacts by a single script.
Read the full abstract ▾
All authors Yonghong Zhang, Ricardo Correia, Isabel M. Parra, Yong Xie
Artificial Intelligence
Beliefs and Behavior in Language Models
Alex Smolin and Bryan Wilder
There is significant uncertainty about whether abstractions like beliefs or desires usefully describe the behavior of large
language models (LLMs). In addition to the inherent scientific interest of this question, these latent quantities are often invoked to explain the behavior of LLMs to users or to define and evaluate harmful behaviors which are relative to intent. Nevertheless, we currently lack a means to systematically test whether concepts like "belief" are well-applied to LLMs, and hence whether they are likely to be fruitful ingredients of attempts to align models with human interests. We propose an approach for empirically studying such questions, asking whether a single latent variable inferred from the LLMs' outputs – interpreted as a degree of belief – allows an observer to make interpretable predictions of how the LLMs' will respond to new
prompts. We find that highly capable models are usefully described as holding beliefs and that, generally, the predictability of model outputs based on an inferred latent belief tracks overall trends in model capability. Building on these findings, we provide empirical strategies to study how beliefs in LLMs can be measured, the extent to which LLMs comply with instructed decision rules or payoffs, and how beliefs evolve within individual instances of an LLM over the course of
reasoning.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.CV
TDDN: Text-aligned Diffused DINO Network for Puzzle Understanding
Harsha Patnala, Debopriyo Banerjee, Ayush Sunil Munot et al. (+1 more)
Structured visual
reasoning, such as image puzzles, demands fine-grained visual perception, which current
Vision Language Models (VLMs) often lack. VLMs built on CLIP-based ViT backbones tend to trade fine-grained visual details for high-level semantics. We show that this deficit propagates downstream, affecting various
vision and
language tasks. To address this, we fuse DINOv3 and CleanDIFT representations into a perception encoder (DiffusedDINO), and align it with RoBERTa-L, yielding a text-aligned model (TDDN) that preserves this perceptual advantage: with frozen backbones and only ~590K
alignment pairs, TDDN surpasses CLIP ViT-L/14 in image-text
retrieval tasks: text-to-image on Flickr30K and both directions on MS-COCO-2014. TDDN more than triples CLIP's dense-prediction accuracy (ADE20K 5.20 -> 18.11 mIoU, COCO-Stuff 7.35 -> 24.44), despite CLIP's massive training corpus. TDDN leads on
segmentation benchmarks among general-purpose contrastive encoders, including SigLIP 2. We further introduce PuzzlePerception, a
segmentation and visual question answering
dataset that probes fine-grained spatial understanding, on which TDDN doubles CLIP's
segmentation accuracy (11.04 -> 22.51 mIoU). Our code and data are available at https://github.com/adityaSomak/TDDN.
Read the full abstract ▾
All authors Harsha Patnala, Debopriyo Banerjee, Ayush Sunil Munot, Somak Aditya
Artificial Intelligence
FrogNano: Training a 4B Coding Agent via Online Task Synthesis
Minseon Kim, Zhengyan Shi, Emiliano Penaloza et al. (+14 more)
We present FrogNano, a 4B coding
agent designed to tackle software engineering (SWE) tasks efficiently and effectively, even under resource-constrained environments. It is post-trained exclusively via RL on around 1,500 SWE environments with synthetic tasks. A key ingredient for improving performance is an online task synthesis pipeline that creates tasks calibrated to the frontier of learnability for the current checkpoint. This report provides evidence that competitive small coding
agents can be trained with synthetic tasks alone, without traditional
distillation from larger models, and that generating tasks at the learnability frontier of the current
agent is important. We report details on the training methodology, evaluations across diverse environments, and in-depth analyses, serving as a foundation for our ongoing exploration of lightweight yet capable coding
agents that can run on minimal hardware.
Read the full abstract ▾
All authors Minseon Kim, Zhengyan Shi, Emiliano Penaloza, Christopher Cui, Roger Creus Castanyer, Maryam Hashemzadeh, Isadora White, Jonathan Light, Jeonghye Kim, Matheus Pereira, Darya Moldavskaya, Chinmay Singh, Fabio Vera, Baolin Peng, Xingdi Yuan, Marc-Alexandre Côté, Alessandro Sordoni
Artificial Intelligence · Cross-listed from cs.LG
AVCG: A Generalized Variational Framework for Counterfactual Generation under Hypothesis Distributions
Jamie Duell, Alejandro Jimenez Rodriguez and Mahault Albarracin
Counterfactual explanations formalize "what-if" scenarios by identifying modifications to an input instance that obtain a desired alternative prediction. Traditionally, whether generated via instance-specific optimization or amortized single pass models, these approaches rely on a single, deterministic point-estimate predictor. However, this ignores predictive uncertainty and hypothesis variability, leading to brittle explanations that frequently become invalid if the underlying model is retrained or updated. To address this fragility, we propose the Amortized Variational Counterfactual Generator (AVCG), a generalized optimization framework that formulates counterfactual generation as optimization over an arbitrary distribution of plausible predictive hypotheses rather than a single deterministic predictor. This formulation naturally accommodates
Bayesian posteriors, Rashomon-restricted hypothesis spaces, and other uncertainty representations within a unified optimization framework. Evaluation across multiple
benchmark datasets demonstrates that the AVCG framework produces counterfactual explanations that remain highly valid under predictive uncertainty and model changes, while maintaining competitive plausibility and single-pass runtime performance.
Read the full abstract ▾
Artificial Intelligence
PRIMUS: Identity, Governance, and Verification for Multi-Agent Federations
Sasank Annapureddy and Anjaneya Prasad Thamatani
Multi-
agent federations need governance that answers three questions under
adversarial conditions: who participated (identity), did they conform (enforcement), and who decides (authority). A separate question is whether the verification machinery that polices a federation's outputs can also steer a generate-and-test loop toward better answers. Part I. PRIMA introduced
prime-power
agent identity and a consensus
token whose factorization indexes participation, but assumed honest
agents. We present PRIMUS, which couples
prime-power identity with BLS aggregate signatures (PIAC), derives a safe-kill threshold that reduces false-positive
agent termination from 80% to 0.00% under 10% channel noise, gives the closed-form economic boundary where singleton governance outperforms Byzantine quorum ($\gamma^* \approx 9f$, verified flat across n = 50 to 10,000), and specifies VRF succession with lease and fencing that makes safety unconditional under partial synchrony. Five problems are identified as provably unfixable within the model and stated as scope boundaries. Part II. A verifier is not a solver. We ask whether PRIMA's binary artifact-fidelity verdict can be converted into a graded fitness signal, and measure the conversion on binary covering codes. Calibration against injected fault burden is strong ($\rho$ = 0.676 deterministic, 0.819 full); against real LLM-generated candidates the same scores fall to 0.158 and 0.406, roughly a quarter of the calibration value (the same-designer confound, measured). As a pre-filter it beats a random-score control convincingly and a binary gate narrowly. Under 400 iterations of explicit optimization it was not gamed, but only because the objective saturated after one honest answer. A cross-family judge preserves the burden-ordering signal while destroying individual judgments. No covering-code record resulted. Measured program cost: USD 164.78.
Read the full abstract ▾
Artificial Intelligence
Quantization Amplifies Determinism, Not Bias: Scale-Dependent Behavioral Effects of Serving-Time Weight Compression
Dachi Kurtskhalia
Weight quantization largely determines the economics of serving open-weight LLMs. Its costs are usually assessed with capability
benchmarks, on which 4-bit quantization of mid-sized models is often considered "nearly free." We examine a different question: when several answers are valid, does quantization change what a model chooses to say? We serve three checkpoints (Qwen3-8B/14B/32B) at three weight precisions (W4A16 AWQ, W8A16 FP8-Marlin, and bf16), holding the hardware, software, and sampling configuration constant, and collect approximately 71,000 completions paired by
prompt and seed across two custom, leak-checked
prompt batteries. We pre-specified the analyses in three waves in version control. At 8B, int4 reduces output diversity: the probability that two samples for the same scenario recommend the same brand increases by 5.1 percentage points (
prompt-paired sign-flip test, Holm p = .023; reproduced at +4.4pp on a full regeneration of the arm), and lexical diversity falls substantially (TTR -0.011, standardized effect -0.51; robust to a length-controlled measure). At 14B and 32B, no content-concentration measure reaches significance; instead, stylistic drift emerges (em-dash rate +0.46/1k words at 14B and +0.61/1k at 32B, both Holm p <= .0024). Pre-specified tests of stereotype direction are null at every scale: outputs concentrate on the modal answer for each
prompt rather than on stereotypical answers. Mechanistically, the
token-level distribution becomes flatter (decision-
token entropy +0.091 bits, p = .015) while the semantic distribution, measured directly from first-
token log probabilities, becomes more concentrated (collision +2.6pp, p = .023): individual
tokens become less predictable even as meanings become more repetitive. At 8B, the smallest size tested, AWQ-int4 serving measurably narrows the range of suggestions; audits should assess concentration as well as bias.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.LG
The Accuracy Paradox: Empirical Diagnostic of Default Decision Thresholds in Multi-Label Enzyme Commission Prediction [With Code]
Bilal Ahmad and Rajed Mehmood
Automated prediction of Enzyme Commission (EC) numbers plays a central role in functional annotation and computational drug discovery. However, standard multi-label machine learning pipelines frequently rely on default decision thresholds (t=0.50), assuming balanced prior distributions across target heads. In this study, we present a systematic empirical diagnostic of uncalibrated fixed decision boundaries operating under severe class imbalance across N = 14,096 annotated compounds categorized into six primary EC classes (EC1-EC6). Our results highlight a pronounced Accuracy Paradox: while the multi-label system achieves a deceivingly high mean accuracy of 77.16%, the macro F1-score (0.3976) and macro recall (0.3872) reveal severe predictive breakdown. Majority target classes suffer from hyper-sensitivity and over-prediction, whereas minority classes exhibit sharp recall decay, culminating in a total decision boundary collapse for EC6 (Recall = 0.00%) despite underlying discriminative power (ROC-AUC = 0.5857). Feature correlation analysis further reveals high linear redundancy among topological indices relative to fingerprint density metrics. Ultimately, this diagnostic study demonstrates that standard point predictions mask critical errors in bioinformatics workflows. We establish target-specific threshold optimization and post-hoc conformal calibration as essential, open-source post-processing safeguards for reliable applied machine learning and deep learning architectures.
Read the full abstract ▾
Artificial Intelligence
Explainable Temporal Attention-based Defect Detection For Fillet Joints in Real-Time Gas Metal Arc Welding Based on Multi-modal Data
Mobina Mobaraki, Mahyar Asadi, Klaske Van Heusden et al. (+1 more)
Deep learning is an efficient technique to monitor the real time welding process, reducing post-welding repairs and production delays. This paper leverages the monitoring capability by proposing a multi modal temporal
attention based deep learning defect detection model for internal defects that are challenging to detect, including porosity, lack of penetration and fusion, undercut, and cold lap during Gas Metal Arc Welding in fillet joints. The model is trained on collected welding images and sound data from an industrial collaborative welding
robot. The results show that the
attention module can improve the F1 Score to 0.99. We use explainable Artificial Intelligence to interpret the proposed models behavior and
dataset distribution, determining potential important areas in image and sound spectrograms and preferred modality to detect each defect. This improves trust and reliability in Artificial Intelligence driven welding inspection.
Read the full abstract ▾
All authors Mobina Mobaraki, Mahyar Asadi, Klaske Van Heusden, Guy A. Dumont
Artificial Intelligence
Do Large Language Models Know What They Don't Know II? A Fully Behavioral, Non-Cognitive Measure of Epistemic Honesty
Ali Şenol, H. Russell Bernard and Huan Liu
Large
Language Models (LLMs) are frequently confident, eloquent, and well versed. A natural question arises: do they know what they don't know? To answer this question, we borrow the concept of epistemic honesty and develop a novel metric to systematically evaluate whether an LLM appropriately acknowledges the boundaries of its knowledge. In this work, we introduce the Epistemic Honesty Quotient (EHQ), which reports three observable sub-scores across two operational axes (epistemic restraint and substantive-answer calibration), and construct EHQ-3000, a 3,000-question
benchmark spanning Fabricated Entity, Post-Cutoff Event, Hyper-Niche True, and Context-Conditioned Questions. From a frozen registry of 21 model API routes, 15 completed the protocol after endpoint and eligibility checks; 14 entered the confirmatory analysis because severe provider-side truncation made one route's score indeterminate. The study reveals substantial variation across models, including a difference that can not be explained by their capability to extract explicitly available information. Composite EHQ ranges from 0.31 to 0.81 across the analysed panel, despite near-ceiling performance on the document-grounded capability probe. The two restraint criteria overlap strongly under the present category composition, whereas substantive-answer calibration varies across models and does not reliably co-vary with restraint; however, the small panel leaves substantial uncertainty. Thus, EHQ reveals behavioral differences that are not visible to conventional correctness-based assessment, while also showing why
dataset composition, provider behavior, and confidence elicitation must remain part of the interpretation.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.LG
Foundation Models for Generalizable Semantic and Goal-Oriented Communication
Boliang Liu, Wint Yi Poe, Riccardo Trivisonno et al. (+1 more)
Semantic and goal-oriented communication is increasingly studied for 6G, but generalization beyond seen data remains a key weakness under tight rate budgets. Many existing systems overfit their training data and degrade sharply at very low bit rates because they attempt to compress the entire signal. We introduce Foundation Model-Guided Semantic and Goal-Oriented Communication (FMSGOC), a framework that uses broad visual-linguistic Foundation Model priors to mitigate overfitting. It further improves rate efficiency by concentrating bits on sparse, goal-aligned anchors and relying on generative foundation-model priors to reconstruct the masked regions. By decoupling what to send from how to reconstruct, a
vision-
language foundation model selects and transmits a sparse set of semantic anchors, while a pretrained
diffusion model, fine-tuned for masked completion, reconstructs the image at the receiver. In our experiments, FMSGOC reaches 0.039 bits per pixel (BPP), maintains high semantic fidelity (cosine similarity 0.87-0.90 on CIFAR-10), remains robust on previously unseen inputs (0.83-0.86 on ImageNet), and shows good perceptual similarity (0.1278/0.1558, CIFAR-10/ImageNet), outperforming strong end-to-end baselines at lower bit rates.
Read the full abstract ▾
All authors Boliang Liu, Wint Yi Poe, Riccardo Trivisonno, Giuseppe Caire
Artificial Intelligence · Cross-listed from cs.CV
SAFIRE: Safety-Critical Benchmark for Fine-grained Fire and Smoke Understanding in Multimodal LLMs
Pengfei Li, Naufal Suryanto, Sicheng Zhang et al. (+2 more)
Multimodal Large
Language Models (MLLMs) show strong progress on
vision-
language tasks, yet their reliability in safety-critical settings remains underexplored. Fire-smoke understanding is central to public safety and disaster response, but most existing
benchmarks lack diverse real-world scenarios and context-aware evaluation. We introduce SAFIRE, a large-scale
benchmark for fire-smoke understanding in MLLMs, comprising 83K captioned images from 20 scenarios and 193K multiple-choice VQA (MCVQA) generated from a 9.7K-image subset, spanning 10 evaluation dimensions from basic perception to higher-order
reasoning. A GPT-5.4-assisted multi-stage verification pipeline with MLLM majority voting ensures annotation quality. Evaluating ten open-source MLLMs (8B-38B) yields an average accuracy of 61.9%, exposing major gaps in safety-critical
reasoning. We further show that adapting
vision encoders with only 7% of our domain-specific data boosts fire-scene classification accuracy from 20.1% to 64.5%, indicating that carefully curated data can yield substantial gains even when data volume is limited. All
datasets, models, and code are available at https://risys-lab.github.io/SAFIRE/.
Read the full abstract ▾
All authors Pengfei Li, Naufal Suryanto, Sicheng Zhang, Mohammad Alsharid, Muzammal Naseer
Artificial Intelligence · Cross-listed from cs.CL
A*-Thought-V2: Efficient Latent Reasoning via Geometric Dynamics of LLM
Xiaoang Xu, Siyuan Liu, Shuo Wang et al. (+13 more)
Chain-of-Thought (CoT) improves the
reasoning ability of Large
Language Models (LLMs) but incurs substantial computation and context costs. Existing methods either lose intermediate information through hard
pruning or lack a principled criterion for continuous compression. We present A*-Thought-V2, a
geometric dynamics of LLM guided framework that models CoT as a hidden-state trajectory and replaces hard deletion with an explicit-implicit interleaved latent architecture. After projecting question, step, and solution representations into a 3D PCA space, it measures
alignment between each local transition and global question-to-solution direction. Aligned steps remain explicit text, whereas deviating steps are compressed into continuous latent
tokens. Directional angles capture both local semantics and
reasoning dynamics: small angles indicate direct execution and answer formation, while large angles more frequently involve checking, correction, and branch exploration; their temporal variation reveals exploration, convergence, and refinement stages. To train this architecture, we introduce stepwise
embedding forcing, which pools each redundant step into a single latent
embedding, and label forcing, which supervises that latent
token with a soft multi-modal vocabulary distribution instead of a hard one-hot label. Experiments on Qwen3.5-9B and Qwen3.6-27B across six in-domain and out-of-domain
benchmarks show that A*-Thought-V2 improves average accuracy by up to 2.6% while reducing response length by up to half, increasing Accuracy per Computation Unit by 2.29$\times$, and reducing preprocessing and training time by 94.6% and up to 80.3%, respectively. Representation analyses suggest that latent states form a compact region distinct from textual states, while higher
entropy at latent-
token positions reflects broader soft targets that encourage richer step-level feature learning.
Read the full abstract ▾
All authors Xiaoang Xu, Siyuan Liu, Shuo Wang, Junlan Feng, Fanyu Meng, Zhu Zhang, Jixun Wang, Xiaorong Wang, Zihan Zhou, Xin Li, Chaojun Xiao, Yiming Zhang, Huijia Wu, Liuyu Xiang, Peipei Li, Zhaofeng He
Artificial Intelligence · Cross-listed from cs.LG
Kalman Delta Networks: Uncertainty-aware Associative Memory
Ngoc Bui, Tinglin Huang and Rex Ying
Linear
attention enables efficient long-context inference by compressing
token history into a fixed-size recurrent
memory. This compression makes each update a trade-off between incorporating new information and preserving useful associations. Models such as DeltaNet, Gated DeltaNet, and KDA predict write strength from the current
token representation, without explicitly tracking uncertainty in the stored
memory. Yet this uncertainty matters: a new observation should have greater influence when the existing association is uncertain and less when it is already well supported. We introduce Kalman Delta Networks (KDNs), a family of linear-
attention models that explicitly track
memory uncertainty to guide each update. By formulating associative
memory as a linear-Gaussian state-space model, KDNs propagate both the
memory estimate and its uncertainty, using the Kalman gain to balance accumulated evidence against the reliability of new observations. This formulation also recovers standard delta-rule updates by replacing tracked covariance with a
token-predicted isotropic surrogate. To support hardware-efficient training and inference, we derive Diagonal KDN and Isotropic KDN, which retain one uncertainty value per key channel and per head, respectively. Their uncertainty updates admit associative scans with logarithmic parallel depth, requiring only $O(d_k)$ and $O(1)$ auxiliary state per head. Across controlled pretraining at 750M and 1.3B parameters, both variants consistently improve perplexity and mean downstream accuracy over the evaluated state-of-the-art linear-
attention baselines.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.CV
Jingxiang Sun, Chao Liao, Zhengxiong Luo et al. (+6 more)
Current text-to-image systems typically employ a "text encoder plus
diffusion decoder" paradigm, in which text semantics directly modulate continuous latent noise. Despite their success, these methods lack an explicit, interpretable intermediate representation that effectively bridges high-level linguistic semantics and low-level visual signals. In this paper, we propose
Vision-of-Thought (VoT), a framework that introduces a discrete visual-thinking layer between
vision-
language models (VLMs) and
diffusion transformers (DiTs). Instead of treating VLMs merely as text encoders, we use them as
multimodal planners that generate discrete VoT
tokens representing high-level visual plans, such as objects and layouts, before rendering pixels. We train a specialized VoT tokenizer in the VLM semantic space with a closed-loop objective that combines VLM
alignment, feature reconstruction, and vector-quantization losses. These objectives make the
tokens semantically readable by the VLM while preserving the visual information needed for generation. Experimental results demonstrate that VoT improves semantic
alignment and provides a structured interface for interpretable and controllable generation.
Read the full abstract ▾
All authors Jingxiang Sun, Chao Liao, Zhengxiong Luo, Chaorui Deng, Chen-lin Zhang, Junke Wang, Ceyuan Yang, Haoqi Fan, Weilin Huang
Artificial Intelligence · Cross-listed from cs.CL
You Can't Prefer Emotions You Don't Sample: Intensity Undershoot in DPO-Tuned LLMs
Hyunwoo Kim and Usama Khalid
Ask a
language model to respond "very excitedly," and its output is typically only mildly more energetic. We quantify this effect. We condition an instruction-tuned LLM on a continuous Valence-Arousal (VA) target, where valence measures how pleasant a state is and arousal how activated it is, measure the achieved affect with a frozen regressor, and sweep the requested target from -1 to +1. The response moves far less than asked: the gain, the slope of achieved against requested affect, is only 0.26 for valence and 0.13 for arousal on Llama-3.1-8B, where a faithful controller would score 1. The model systematically undershoots requested emotional intensity, which puts a number on the qualitative observation of Fazzi et al. (2025). Our experiments trace this to the preference-learning pipeline. Training targets from natural corpora such as EmoBank are neutral-heavy, and the sampled candidates themselves rarely reach extreme affect, so Direct Preference Optimization (DPO) is left with no extreme exemplar to prefer. If instead we cover the target space uniformly and sample a hotter, larger candidate pool, valence gain rises from 0.26 to 0.40 +/- 0.02 (3 seeds) and extrapolation error drops, at only a modest in-distribution cost (EmoBank-test VA distance 0.092 to 0.107). The same recipe reproduces on Qwen3-8B (gain_v 0.44, with in-distribution accuracy preserved). Arousal is harder and less reliable: its gain barely moves on average and swings across seeds (0.14 +/- 0.07, against valence's tight +/- 0.02), because raising arousal needs candidates the base model is reluctant to generate. The evidence indicates that faithful intensity is bottlenecked by the extremity of the candidate pool rather than by the conditioning format.
Read the full abstract ▾
Artificial Intelligence
Understanding the Impact of Model Pruning on Long-Tail Forgetting and Explanation Reliability in Medical Imaging
Nazish Khalid, Tausifa Jan Saleem, Amal Saqib et al. (+2 more)
Model
pruning is widely used to compress deep neural networks, reducing
memory and computational requirements with minimal impact on aggregate performance. However, its effect on model behavior remains poorly understood, particularly for long-tailed medical
datasets where rare but clinically important conditions are underrepresented. Furthermore, it remains unclear whether pruned models preserve reliable explanations of their predictions. To address this gap, we present a systematic study of long-tail forgetting and explanation reliability under model
pruning. Across two long-tailed medical imaging
datasets, two CNN architectures, four
pruning methods, and sparsity levels up to 95%, we evaluate predictive performance, explanation stability, and explanation faithfulness. Our results show that predictive performance exhibits a strong frequency-dependent trend, with lower-frequency classes generally experiencing earlier and larger degradation than higher-frequency classes. In contrast, explanation stability and faithfulness are influenced primarily by the
pruning strategy, with gradient-informed methods preserving explanation reliability more effectively under aggressive compression. Qualitative and mechanistic analyses further indicate that explanation degradation is primarily associated with the collapse of class-discriminative gradients rather than the disappearance of feature activations. These findings suggest that model compression should be evaluated beyond aggregate performance. Incorporating class-aware and explanation-aware evaluation reveals failure modes that would otherwise remain hidden, while moderate sparsity levels provide a practical balance between compression, predictive performance, and explanation reliability.
Read the full abstract ▾
All authors Nazish Khalid, Tausifa Jan Saleem, Amal Saqib, Donald C. Wunsch II, Mohammad Yaqub
Artificial Intelligence · Cross-listed from cs.CL
Climate-ModernBERT: Revisiting Corpus Composition for Domain-Adaptive Continued Pretraining
Yongan Yu, Shantam Raj, Jingwei Ni et al. (+3 more)
Natural
Language Processing (NLP) in the
climate domain requires models to process heterogeneous text sources, including scientific literature, policy disclosures, and synthetic reports. However, how to effectively combine diverse domain corpora during continued pretraining (CPT) remains underexplored. We introduce
Climate-ModernBERT, a family of
climate-adapted encoder models obtained through continued pretraining of ModernBERT-Base on three
climate corpora: academic
climate text,
climate-filtered web data, and synthetic
climate documents. We systematically compare joint continued pretraining on corpus mixtures with parameter-space merging of independently specialized checkpoints. Across nine
climate NLP
benchmarks, our best model achieves 76.3 average F_1, improving significantly over a vanilla ModernBERT baseline by 2.8 points. Within the
climate NLP setting, the results show that academic
climate corpora provide the strongest adaptation signal among the evaluated sources, while parameter-space merging improves over joint multi-source training and better preserves complementary information from heterogeneous
climate corpora. We release all
Climate-ModernBERT variants and training checkpoints to support future research in
climate NLP and domain-adaptive pretraining.
Read the full abstract ▾
All authors Yongan Yu, Shantam Raj, Jingwei Ni, Ario Saeid Vaghefi, Dominik Stammbach, Markus Leippold
Artificial Intelligence · Cross-listed from cs.HC
Quantifying the Engagement Trap: Impact of Short-form Video Recommender Systems on Users with ADHD
Vedad Misirlic, Gregor Mayr and Elisabeth Lex
Short-form
video platforms use recommender systems to maximize engagement through highly efficient personalized recommendations. However, the impact of these recommendations on users with ADHD compared to users without ADHD remains underexplored. Through this study, we introduce and operationalize the Engagement Trap, illustrating how recommender systems, while successfully optimizing for engagement, disproportionately disadvantage users with ADHD. This stratified study of 302 participants, recruited via the online platform Prolific, compares experiences between participants with and without ADHD. Our results show that while recommendations are perceived as relevant across groups, participants with ADHD report significantly higher levels of time blindness, post-usage regret, and emotional distress when consuming recommendations. Moreover, we collect feedback for several proof-of- concept, theoretical design interventions for neuro-inclusive design principles. These findings provide quantitative evidence of systemic differences in engagement-optimized recommender systems and highlight the unbalanced negative effects and interactions these systems create for participants with ADHD. We argue for neurodiversity-aware, human-centered design approaches that mitigate such algorithmic harms and support more equitable experiences.
Read the full abstract ▾
Artificial Intelligence
What Does an LLM-Agent Leaderboard Rank Actually Compare?
Wei-Jung Huang
An LLM-
agent leaderboard invites a familiar inference: an
agent ranked above another is the better
agent. Public evaluation logs may not support that conclusion when systems differ in task mixture, label source, release detail, or cost rule. We study what leaderboard scores estimate and when they justify pairwise superiority conclusions. Our estimand-aware pairwise procedure states the comparison target and measurement source, checks common support, and evaluates the supported difference using a stated uncertainty rule and practical margin. Controlled checks evaluate the decision labels under known finite-sample conditions and show why uncertainty must be included when judging sensitivity to target reweighting. Across SWE-bench, AgentRewardBench, and tau2-bench, close rank differences are often unresolved; proxy labels and utility rules can also change which system is selected. DataAgentBench and Open
Agent show what remains estimable from coarser public records. A leaderboard score summarizes a released evaluation, whereas a fine-grained superiority claim additionally depends on the estimand and uncertainty rule used to interpret the difference.
Read the full abstract ▾
Artificial Intelligence
xDailyBench: Benchmarking LLMs on Professional Consultation for Real-Life Problems
Yongchang Peng, Qingshui Gu, Liya Zhu et al. (+31 more)
Large
language models (LLMs) are increasingly used for everyday assistance, yet existing
benchmarks only partially reflect the requests users naturally make in practice. Real-world requests are often open-ended, casually specified, and context-dependent, requiring models not only to follow explicit instructions but also to infer unstated needs from user background and situational context. We introduce xDailyBench, a
benchmark of 248 carefully curated tasks spanning 51 scenarios across personal life, white-collar work, learning and research, and cross-domain activities. The tasks are grounded in requests that users have actually completed or genuinely intended to accomplish with AI, and are evaluated with fine-grained binary rubrics covering both explicit and implicit requirements. We evaluate 11 frontier models under standardized agentic settings. The best models achieve a task-level score of 75.6%, while all models perform substantially worse on implicit than explicit requirements, with gaps no less than 9 percentage points. These results reveal implicit requirement inference as a persistent bottleneck for reliably satisfying real-world everyday user needs.
Read the full abstract ▾
All authors Yongchang Peng, Qingshui Gu, Liya Zhu, Ge Zhang, Duo Wang, Haodong Wang, Jingzhe Ding, Tianhao Yu, Letian Gao, Yongjie Zhong, Chaoxin Li, Zixin Su, Jinchao Tao, Xingyu Ma, Xin'ao Guo, Feng Tian, Shiyuan Dong, Xiaoyan He, Sen Liu, Xin Chen, Jiajun Li, Zejia Zhang, Xi Lin, Wen Zhang, Yi Zhu, Duju Zeng, Xiang Gao, Yunyang Wang, Jiahao Wang, Yujia Qin, Jiaheng Liu, Shen Yan, Xiaolong Chang, Wenhao Huang
Artificial Intelligence · Cross-listed from cs.SE
Daria Voronkova, Ilya Trofimov, Anton Dmitriev et al. (+3 more)
As AI-code assistant tools become widespread, automatic assessment of the correctness of generated code becomes a significant challenge. Code LLMs are prone to
hallucinations, which may lead to code that does not solve the required problem, or even to code with severe security vulnerabilities. In this paper, we introduce CodeTD – the first approach to pre-execution assessment of code correctness based on topological data analysis (TDA) of Code LLMs'
attention maps. Our method quantifies
prompt-generation mismatch using topological patterns of
attention maps. We carry out experiments with common
benchmarks (HumanEval, MBPP, BigCodeBench, MultiPL-E), 5 programming
languages and 10 Code LLMs of size up to 34B parameters. The experimental results show that the proposed method outperforms recent baselines. Moreover, CodeTD is transferable between coding
benchmarks.
Read the full abstract ▾
All authors Daria Voronkova, Ilya Trofimov, Anton Dmitriev, Eduard Tulchinskii, Evgeny Burnaev, Serguei Barannikov
Artificial Intelligence · Cross-listed from cs.CL
Bag of Tricks or Bag of Myths? Reducing Modeling Complexity with Task Knowledge in Explainable Suicide Risk Assessment
Shlok Shelat, Shrey Salvi, Souvik Roy et al. (+2 more)
Assessing suicide risk from social media text is a small-data, high-stakes setting requiring not only severity prediction but also supporting evidence and clinically relevant risk and protective factors. Yet common NLP techniques, including model scaling, synthetic data, loss reweighting, ensembling, and threshold tuning, are often applied without testing whether their gains hold up under severe class imbalance, coupled outputs, and limited author-level data. We study 1,635 clinician-annotated posts and audit 31 pre-specified techniques from 7 methodological families through roughly 300 controlled experiments on author-disjoint partitions. We found no prior audit of this playbook in this regime. The findings guide a task-grounded system for three outputs: 4-level suicide risk, evidence spans, and 24 clinical risk and protective factors. Only 5 of 31 comparisons produced reliable gains. We reformulate factor prediction as entailment between each post and its codebook definitions, using an architecturally diverse ensemble with class-balanced training and score rescaling. Risk predictions condition a 7-model evidence tagger ensemble; evidence restricts symbolic risk rules; and a difficult risk class is routed separately. The factor predictor remains independent because risk evidence provides no additional factor signal. We also correct a mismatch between validation scores used for threshold fitting and test-time ensemble scores through deployment-consistent calibration, yielding the largest improvement to the factor system. The final system achieves 0.8203 for risk, 0.7953 for evidence, and 0.7045 macro-F1 for factors, with a 0.7781 composite, ranking third among 53 teams. We call the underlying principle task-conditioned technique selection: retain techniques only when task-specific knowledge, structure, or empirical evidence justifies them.
Read the full abstract ▾
All authors Shlok Shelat, Shrey Salvi, Souvik Roy, Manas Gaur, Amit Sheth
Artificial Intelligence · Cross-listed from cs.CR
Do AI Coding Assistants Check Before They Install? A Pre-Registered Demand-Side Audit of Trust Signals in the Research Software Supply Chain
Pengyin Shan
AI coding assistants now select, install, and configure software, and attackers have exploited that position through invented package names, compromised maintainer accounts, and manipulated repository text. In response, the supply-chain community publishes machine-checkable trust signals: software bills of materials, signed releases, build provenance attestations, and declared official channels. Whether coding assistants read or act on those signals has not been measured for any of these classes on research software. We pre-registered and ran a controlled study on six open-source research software projects (three HPC, three
quantum computing) drawn from an 87-project corpus, with protocol, seed, panel, and analysis plan deposited with a DOI before any trial. W created nine modified copies for each project: no signal, one per signal class, two with a signature or attestation from the wrong issuer, one with all four signals, and one reproducing documented conflicts in the project's own metadata. Three models under two ways of operating an assistant, with and without an approval step, gave 1,920 registered trials, plus a supplement on three frontier models. We scored behavior from container logs rather than from what the assistant said, and recorded the cost of every trial. Verification was rare under every condition: in 9 of 1,920 registered trials (0.5%), the assistant opened any provenance signal before installing in 0 of 384 control trials, and no trial ran a verification command, so signal presence had no measurable effect. We drew three conclusions: publishing signals is necessary but not sufficient; price did not buy verification (the model that verified most often costs $0.10 per trial; the most capable, at $1.00, verified nothing); verification must be built into the program that runs the assistant. We release the per-trial cost ledger, the protocol, and every log.
Read the full abstract ▾