# Research Papers — 2026-05 · bundle page 234 of 326

> Full-text markdown of papers #11651–11700 of 16266 published in 2026-05, newest first. AI-readable mirrors of the full 16266-paper 2026-05 archive (50 papers per file).

Canonical page: https://aipod.dev/papers/archive/2026-05/page/234
[← Previous bundle](https://aipod.dev/papers/archive/2026-05/page/233.md)
[Next bundle →](https://aipod.dev/papers/archive/2026-05/page/235.md)

### Quantifying the Utility of User Simulators for Building Collaborative LLM Assistants

Authors: Joseph Suh, Ayush Raj, Minwoo Kang, Serina Chang
Published: 2026-05-10
Categories: cs.CL
arXiv: https://arxiv.org/abs/2605.09808v1

User simulators are increasingly leveraged to build interactive AI assistants, yet how to measure the quality of these simulators remains an open question. In this work, we show how simulator quality can be quantified in terms of its downstream utility: how an LLM assistant trained with this user simulator performs in the wild when interacting with real humans. In a controlled experiment where only the user simulator varies, we train LLM assistants via reinforcement learning against a spectrum of simulators, from an LLM prompted to role-play a user to one fine-tuned on human utterances from WildChat. As evaluation, we measure pairwise win rates in a user study with 283 participants and on WildBench, a benchmark derived from real human--AI conversations. Training against the role-playing LLM yields an assistant statistically indistinguishable from the initial assistant in our user study (51% win rate), whereas training against the fine-tuned simulator yields significant gains (58% over the initial and 57% over the one trained against role-playing). Closer inspection reveals three further patterns: methods for making role-playing LLMs more realistic (e.g., persona conditioning) improve trained assistants but do not close the gap to the fine-tuned simulator; scaling the simulator's model size benefits the fine-tuned simulator but yields no gain for role-playing ones; and assistants trained against role-playing simulators fail to generalize when paired with other simulators at test time, while the one trained against fine-tuned simulator does. Together, these results argue for grounding user simulators in real human behavior and measuring their quality by their downstream effect on real users.

---

### LEAD: Length-Efficient Adaptive and Dynamic Reasoning for Large Language Models

Authors: Songtao Wei, Yi Li, Zhikai Li, Xu Hu, Yuede Ji, Guanpeng Li, Feng Chen, Carl Yang, Zhichun Guo, Bingzhe Li
Published: 2026-05-10
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2605.09806v1

Large reasoning models, such as OpenAI o1 and DeepSeek-R1, tend to become increasingly verbose as their reasoning capabilities improve. These inflated Chain-of-Thought (CoT) trajectories often exceed what the underlying problems require, wasting compute, latency, and context budgets. While introducing length-based efficiency rewards during reinforcement learning offers a natural remedy, existing methods struggle with two fundamental challenges: the optimal balance between correctness and efficiency is non-stationary throughout training, and intrinsic reasoning budgets vary drastically across problems. Relying on static reward weights and global length constraints inevitably forces a compromise between degraded accuracy and unrealized compression. To overcome these limitations, we propose LEAD (Length-Efficient Adaptive and Dynamic reasoning), a method that replaces static heuristics with online, self-adaptive mechanisms. LEAD dynamically calibrates the correctness-efficiency trade-off at each step using a Potential-Scaled Instability, directing optimization capacity to the most informative learning signal. Furthermore, it estimates an adaptive per-problem target length online based on the model's own correct rollouts, applying a symmetric efficiency reward that penalizes both overthinking and over-compression. Evaluated on five mathematical reasoning benchmarks, LEAD achieves the highest accuracy and Accuracy-Efficiency Score among RL-trained efficient-reasoning methods while producing substantially shorter outputs than the base model.

---

### Insight: Enhancing Mobile Accessibility for Blind and Visually Impaired Users with LLMs

Authors: Joshua Owusu Ansah, Anuj Kapoor, Ayush Khanna, Manvika Vinod, Precious Njeck, Shuai Gao
Published: 2026-05-10
Categories: cs.HC, cs.AI
arXiv: https://arxiv.org/abs/2605.09803v1

This research paper addresses the limitations of current mobile accessibility services like TalkBack, which provide manual gesture-based sequential feedback to BVI users. Motivated by the promise of large language models (LLMs), this paper introduces Insight, an Android accessibility service that provides natural language interaction and real-time summarization of the screen. The paper performs a within-subject experimental study with users to compare Insight and TalkBack on usability factors. Results show Insight reduced mental effort and task time, and was preferred because of its dialogue interface, but users felt the need for interruption management. Results show LLM-based interfaces can significantly improve mobile accessibility, and describe the potential of hybrid solutions combining gesture and dialogue modalities towards more inclusive design.

---

### CrossVL: Complexity-Aware Feature Routing and Paired Curriculum for Cross-View Vision-Language Detection

Authors: Zhipeng Liu, Chunbo Luo
Published: 2026-05-10
Categories: cs.CV, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2605.09802v1

Vision-language models (VLMs) enable text-guided object detection but degrade severely under cross-view scenarios where ground and aerial viewpoints differ in altitude, scale, and spatial layout. These geometric changes introduce systematic complexity variations between viewpoints, e.g., ground view images contain dense and highly occluded structures, while aerial images are sparse and globally organized. Fixed VLM fusion mechanisms cannot handle this discrepancy. We propose CrossVL, a framework combining Complexity-Aware Pathway Aggregation (CPA) and Paired Curriculum Learning (PCL) for enhanced cross-view detection for VLM. CPA estimates scene complexity from multimodal statistics and routes visual features through multiple pathways to obtain view-specific representations. PCL leverages semantic consistency of synchronized ground-aerial pairs to provide stable early supervision and then gradually shifts toward randomized sampling. On MAVREC, CrossVL improves Florence-2's aerial mAP from 58.66% to 61.03% and reduces the ground-aerial performance gap from 8.63pp to 6.65pp, while also achieving a 3.3x reduction in variance across random seeds. CPA provides stable complexity-aware feature aggregation, and PCL enhances optimization dynamics. Together, they demonstrate that coordinated architectural and training adaptations are crucial for robust cross-view VLM detection.

---

### cantnlp@DravidianLangTech 2026: organic domain adaptation improves multi-class hope speech detection in Tulu

Authors: Andrew Li, Sidney Wong
Published: 2026-05-10
Categories: cs.CL
arXiv: https://arxiv.org/abs/2605.09795v1

This paper presents our systems and results for the Hope Speech Detection in Code-Mixed Tulu Language shared task at the Sixth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages (DravidianLangTech-2026). We trained an XLM-RoBERTa-based text classification system for detecting hope speech in code-mixed Tulu social media comments. We compared this organically adapted hope speech detection model with our baseline model. On the development set, the organically adapted model outperformed the baseline system. While our submitted systems performed more modestly on the official test set, these results suggest that further adapting XLM-RoBERTa on organically collected Tulu social media text containing code-mixed and mixed-script variation can improve hope speech detection in code-mixed Tulu.

---

### Multi-Tier Labeling and Physics-Informed Learning for Orbital Anomaly Detection at Scale

Authors: Yong Fu
Published: 2026-05-10
Categories: cs.DC, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2605.09790v1

Detecting orbital anomalies, such as maneuvers, atmospheric decay, and attitude upsets, across the rapidly growing population of low-Earth-orbit (LEO) satellites is a prerequisite for collision avoidance, decay forecasting, and conjunction screening. The bottleneck is not modeling capacity but labels: there is no public ground-truth corpus of orbital anomalies, manual review does not scale to approximately 10^4 active satellites, and pure rule-based detectors trade recall for precision so aggressively that they are blind to most behavioral anomalies. We present a multi-tier labeling cascade that composes three weak supervision sources of increasing fidelity: a fast physics rule set (rule_v1), an Interacting Multiple Model Unscented Kalman Filter (IMM-UKF) bank, and a supplemental-element calibration step (supGP), to produce labels at a scale unavailable from any single source. Applied to 232M Two-Line Element (TLE) records spanning 60 years, the cascade yields 8.6M labeled sequences of length 50 (430M timesteps) over 11 features that include explicit time encoding and full mean-element state. On overlapping satellites, IMM-UKF surfaces 42.6x more anomalies than rule_v1 alone. We train a 6.5M-parameter Transformer in two stages, achieving a maneuver recall of 55.4% and decay recall of 62.8% on a held-out test set. An ablation on the time-delta feature alone yields a 107% relative improvement in decay recall. We frame the resulting model as a high-recall triage classifier whose role is to surface candidate events for downstream filtering, not to issue final attributions, and discuss the path toward a Neural-ODE-based orbital world model.

---

### Parameter-Efficient Neuroevolution for Diverse LLM Generation: Quality-Diversity Optimization via Prompt Embedding Evolution

Authors: Dongxin Guo, Jikun Wu, Siu Ming Yiu
Published: 2026-05-10
Categories: cs.NE, cs.AI, cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2605.09781v2

Large Language Models exhibit mode collapse, producing homogeneous outputs that fail to explore valid solution spaces. We present QD-LLM, a framework for parameter-efficient neuroevolution that evolves prompt embeddings, compact neural interfaces (~32K parameters) that steer generation in frozen LLMs (70B+ parameters), within a Quality-Diversity (QD) optimization framework. Our contributions: (1) evolved prompt embeddings via gradient-free optimization enabling behavioral steering without model fine-tuning; (2) hybrid behavior characterization combining semantic and explicit features with formal coverage bounds (Theorem 1) under validated near-independence (NMI $= 0.08 \pm 0.02$); (3) co-evolutionary variation operators including targeted behavioral mutation via finite-difference gradient estimation. On HumanEval (164 problems), MBPP, and creative writing benchmarks, QD-LLM achieves 46.4% higher coverage and 41.4% higher QD-Score than QDAIF ($p<0.001$, 30 runs, Vargha-Delaney $A=0.94$). We demonstrate downstream utility: diverse archives improve test generation (34% more edge cases) and fine-tuning data quality (8.3% accuracy gain). We validate across open-source LLMs (Llama-3-70B, Mistral-Large) with full embedding access, establishing prompt embedding evolution as an effective paradigm bridging neuroevolution and modern LLMs.

---

### Attribution-based Explanations for Markov Decision Processes

Authors: Paul Kobialka, Andrea Pferscher, Francesco Leofante, Erika Ábrahám, Silvia Lizeth Tapia Tarifa, Einar Broch Johnsen
Published: 2026-05-10
Categories: cs.AI
arXiv: https://arxiv.org/abs/2605.09780v2

Attribution techniques explain the outcome of an AI model by assigning a numerical score to its inputs. So far, these techniques have mainly focused on attributing importance to static input features at a single point in time, and thus fail to generalize to sequential decision-making settings. This paper fills this gap by introducing techniques to generate attribution-based explanations for Markov Decision Processes (MDPs). We give a formal characterization of what attributions should represent in MDPs, focusing on explanations that assign importance scores to both individual states and execution paths. We show how importance scores can be computed by leveraging techniques for strategy synthesis, enabling the efficient computation of these scores despite the non-determinism inherent in an MDP. We evaluate our approach on five case-studies, demonstrating its utility in providing interpretable insights into the logic of sequential decision-making agents.

---

### Nectar: Neural Estimation of Cached-Token Attention via Regression

Authors: João Monteiro, Michal Klein, Pierre Ablin, Marco Cuturi
Published: 2026-05-10
Categories: cs.LG, cs.CL
arXiv: https://arxiv.org/abs/2605.09778v2

Evaluating softmax attention over a fixed long context requires reading every cached key-value pair for each new query token. For a given context (a book, a manual, a legal corpus) the attention output is a deterministic function of the query. We propose Nectar, which fits a compact neural network to this function for queries drawn from a task-relevant distribution. Nectar fits two networks per layer and KV-head: a target network that predicts the attention output and a score network that predicts the log-normalizer. The pair plugs into the standard masked self-attention at inference time, replacing the $O(n)$ attention over the cache with a forward pass whose cost does not depend on $n$. Each module carries on the order of $|θ|$ parameters per layer and KV-head, typically much smaller than the $2nd$ KV-cache footprint at the same granularity. We report experiments on models from 1.7B to 8B parameters across five long-context datasets. The approximation error tracks the next-token accuracy gap to full attention, and allocating capacity non-uniformly across layers reduces that gap in our ablation. Beyond this analysis of metrics, we check that the text generations (following a question prompt) of a model equipped with a Nectar module match in semantic content those obtained by giving the same model access to the full cache.

---

### EvoPref: Multi-Objective Evolutionary Optimization Discovers Diverse LLM Alignments Beyond Gradient Descent

Authors: Dongxin Guo, Jikun Wu, Siu Ming Yiu
Published: 2026-05-10
Categories: cs.NE, cs.AI, cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2605.09777v2

Gradient-based preference optimization methods for large language model (LLM) alignment suffer from preference collapse, converging to narrow behavioral modes while neglecting preference diversity. We introduce EvoPref, a multi-objective evolutionary algorithm that maintains populations of Low-Rank Adaptation (LoRA) adapters optimized across helpfulness, harmlessness, and honesty objectives using Non-dominated Sorting Genetic Algorithm II (NSGA-II) selection with archive-based diversity preservation. Our primary contribution is demonstrating that population-based methods discover substantially more diverse alignments than gradient descent. On standard benchmarks, EvoPref improves preference coverage by 18% (median 82.5% vs. 70.0% for ORPO, $p<0.001$, Wilcoxon, $n=30$) and reduces collapse rates by 47% (11.0% vs. 20.6%, $p<0.001$), while achieving competitive alignment quality (median 75.5% RewardBench vs. 75.0% for ORPO, $p<0.05$). We provide theoretical motivation extending recent multi-objective evolutionary algorithm (MOEA) runtime analysis (Dang et al., 2025) suggesting why archive-based methods escape collapse more effectively than single-trajectory optimization. Comprehensive comparisons against MOEA/D, SMS-EMOA, CMA-ES, and gradient baselines (DPO, IPO, KTO, ORPO) with rigorous statistical testing (Friedman with Holm correction, Vargha-Delaney effect sizes, median with IQR) confirm that multi-objective selection with diversity preservation is essential. This work establishes evolutionary optimization as a principled paradigm for diverse LLM alignment.

---

### Bayesian Optimization with Structured Measurements: A Vector-Valued RKHS Framework

Authors: Wenbin Wang, Colin N. Jones
Published: 2026-05-10
Categories: cs.LG, math.OC
arXiv: https://arxiv.org/abs/2605.09775v1

Bayesian optimization (BO) is an efficient framework for optimizing expensive black-box functions. However, it is typically formulated as learning an end-to-end mapping from inputs to scalar objectives, thereby discarding the potentially rich information whenever a structured system output is available. In this work, we study Bayesian optimization over a vector-valued operator with structured measurements, where each measurement observes multidimensional or functional outputs, e.g., trajectories or spatial fields, rather than a single scalar value. The objective is then defined as a linear functional of these measurements. This allows each observation to reveal substantially richer information about the underlying system compared to scalar observations. Assuming the unknown operator lies in a vector-valued reproducing kernel Hilbert space (RKHS), we derive high-probability concentration bounds for the kernel ridge regression (KRR) estimator directly in the measurement space, characterizing uncertainty in a general Hilbert space. Building on these results, we propose an algorithm based on the upper confidence bound (UCB) acquisition function with regret guarantees under mild assumptions, recovering sublinear rates for common kernels. Empirically, we demonstrate that leveraging structured measurements leads to improved sample efficiency by enabling efficient transfer of information across objectives and adaptation to time-varying settings.

---

### DRIVE-C: A Controlled Corruption Dataset for Autonomous Driving

Authors: Shiva Aher
Published: 2026-05-10
Categories: cs.CV
arXiv: https://arxiv.org/abs/2605.09774v1

DRIVE-C is a controlled corruption dataset designed to evaluate visual perception robustness in autonomous driving systems. It is built from real-world forward-facing driving videos collected across daytime, nighttime, urban, rural, freeway, and parking environments. Clean clips are anonymized via localized face and license plate blurring, then transformed with physics-inspired synthetic degradations. The dataset contains 10 clean clips and 600 corrupted clips spanning 12 camera degradation types across five severity levels, with per-clip metadata and Global Sensor Health Index (GSHI) annotations. DRIVE-C supports robustness benchmarking, degradation-aware modeling, uncertainty estimation, out-of-distribution (OOD) detection, and sensor health monitoring for Advanced Driver Assistance Systems (ADAS). By providing pixel-aligned clean and degraded video clips with fully reproducible corruption parameters, DRIVE-C offers a structured testbed for studying perception reliability under controlled camera degradation.

---

### Exploitation Without Deception: Dark Triad Feature Steering Reveals Separable Antisocial Circuits in Language Models

Authors: Cameron Berg, Roshni Lulla
Published: 2026-05-10
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2605.09773v1

We use sparse autoencoder (SAE) feature steering to amplify Dark Triad personality traits (Machiavellianism, narcissism, and psychopathy) in Llama-3.3-70B-Instruct and evaluate the resulting behavioral changes across five psychological instruments. The steered model becomes substantially more exploitative, aggressive, and callous on novel behavioral scenarios (d=10.62) while its cognitive empathy remains intact, reproducing the empathy dissociation characteristic of human Dark Triad populations. Critically, strategic deception is completely unaffected across all features, suggesting that exploitation and deception may operate through dissociable computational pathways in large language models. Individual feature analysis reveals non-redundant encoding, with each feature driving distinct antisocial mechanisms through separable computational pathways. We also show that feature discovery method itself modulates intervention depth: contrastively-discovered features change both self-report and behavior, while semantically-searched features change only self-report (d=12.65 between methods on behavior). These findings suggest that antisocial tendencies in at least one large language model comprise dissociable components rather than a unified construct, with implications for how such tendencies should be detected, measured, and controlled.

---

### Marrying Generative Model of Healthcare Events with Digital Twin of Social Determinants of Health for Disease Reasoning

Authors: Ziquan Wei, Tingting Dan, Guorong Wu
Published: 2026-05-10
Categories: cs.AI
arXiv: https://arxiv.org/abs/2605.09771v1

Despite the central role of sensor-derived measurements such as imaging traits and plasma biomarkers in biomedical research and clinical practice, existing generative models for disease prediction largely depend on event-level representations from hospital and registry data. Given the multi-factorial nature of human disease, the absence of explicit modeling of social determinants of health (SDoH), even in the limited form of ICD-coded proxies (chapters Z and V--Y in ICD-10), limits the capacity for personalized disease modeling and clinical decision support. To address this limitation, we propose a generative model with ICD-coded proxies of SDoH for \textit{in silico} modeling of disease reasoning, a conditioned latent diffusion framework that establishes the connection between multi-organ sensor data with tokenized healthcare events. Specifically, we introduce a novel geometric diffusion model to characterize the temporal evolution of complex data representation such as brain networks (region-to-region connectivity encoded in a graph), in parallel with diffusion models for tabular data from other organ systems. Together, we integrate the generative model with digitalized SDoH proxies (coined \modelname{}) for simulated intervention and reasoning of future disease trajectories. We conduct extensive experiments on the UK Biobank (UKB) dataset, which contains organ-specific imaging traits, including brain (44,834), heart (23,987), liver (28,722), and kidney (32,155), along with nearly 500k medical history sequences (age range: 25$\sim$89 years). Our \modelname{} achieves significant improvements over state-of-the-art human disease autoregressive models and imaging trait generative baselines.

---

### Encoding and Decoding Temporal Signals with Spiking Bandpass Wavelets

Authors: Jens Egholm Pedersen, Tony Lindeberg, Peter Gerstoft
Published: 2026-05-10
Categories: cs.NE, eess.SP, q-bio.NC
arXiv: https://arxiv.org/abs/2605.09770v1

Spike-based encodings are sparse and energy-efficient, but have largely been formulated probabilistically, disconnected from most signal processing literature. We recast spike encoders as time-causal wavelet frames with quantitative bandwidths and reconstruction error bounds. The proposed wavelets preserve the sparsity and locality of spiking representations, with reconstruction up to spike quantization and time discretization. We demonstrate reconstruction on ECG and audio datasets, achieving a normalized RMSE comparable to continuous wavelet transforms. The spiking wavelets map directly to neuromorphic hardware.

---

### UTS at PsyDefDetect: Multi-Agent Councils and Absence-Based Reasoning for Defense Mechanism Classification

Authors: Dima Galat, Marian-Andrei Rizoiu
Published: 2026-05-10
Categories: cs.AI
arXiv: https://arxiv.org/abs/2605.09769v2

This paper describes our system for classifying psychological defense mechanisms in emotional support dialogues using the Defense Mechanism Rating Scales (DMRS), placing second (F1 0.406) among 64 teams. A central insight is that defense mechanisms are defined by what is absent: missing affect, blocked cognition, denied reality. We encode this as an affect-cognition integration spectrum in prompt-level clinical rules, which account for the largest single gain (+11.4pp F1). Our architecture is a multi-phase deliberative council of Gemini 2.5 agents where class-specific advocates rate evidence strength rather than voting, achieving F1 0.382 with no fine-tuning - a top-5 result on its own. We find, however, that the council is confidently wrong about minority classes: 59-80% of stable minority predictions are incorrect, driven by a systematic "L7 attractor" in which emotional content defaults to the majority class. A targeted override ensemble from three fine-tuned Qwen3.5 models applies 16 overrides (+2.4pp), selected by a structured multi-agent system (builder, critic, regression guard) that produced a larger F1 gain in one iteration than 8 prior attempts combined.

---

### WISTERIA: Learning Clinical Representations from Noisy Supervision via Multi-View Consistency in Electronic Health Records

Authors: Ruan Dong, Yuanyun Zhang, Shi Li
Published: 2026-05-10
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2605.09765v1

Representation learning in electronic health records (EHR) has largely followed paradigms inherited from natural language processing, relying on sequence modeling and reconstruction based objectives that treat clinical labels as ground truth. However, real world clinical supervision is inherently weak, arising from heterogeneous, noisy, and institution specific labeling processes such as billing codes, heuristic phenotypes, and incomplete annotations. In this work, we propose WISTERIA, a weakly supervised representation learning framework that models labels as stochastic observations of an underlying latent clinical state. Instead of optimizing against a single supervision signal, WISTERIA constructs multiple weak supervision operators and learns representations by enforcing consistency across their induced label distributions. This multi view formulation induces an implicit denoising mechanism, allowing the model to recover clinically meaningful structure by reconciling disagreement between noisy labelers. We further incorporate ontology aware regularization in the label space to impose semantic structure over supervision signals. Empirically, WISTERIA improves predictive performance across standard EHR benchmarks, demonstrates strong robustness to label noise, and exhibits superior cross institutional generalization compared to sequence based pretraining objectives. These results suggest that explicitly modeling the supervision process rather than treating labels as fixed targets provides a more appropriate inductive bias for learning robust and clinically meaningful representations from EHR data.

---

### LEVI: Stronger Search Architectures Can Substitute for Larger LLMs in Evolutionary Search

Authors: Temoor Tanveer
Published: 2026-05-10
Categories: cs.NE, cs.AI
arXiv: https://arxiv.org/abs/2605.09764v1

LLM-guided evolutionary methods such as AlphaEvolve have proven effective in domains like math, systems research, and algorithmic discovery, but their reliance on frontier models makes each run expensive. We argue this is largely an artifact of how existing frameworks allocate search: archives that fail to preserve solution diversity force compensation through stronger mutation models; blind model use spends frontier dollars on local edits a smaller model could handle; and full-set evaluation wastes rollouts on redundant examples. We introduce LEVI, a harness-first evolutionary framework built on the bet that stronger search architectures can substitute for or even outperform larger LLMs in evolutionary search. LEVI improves on three core components of evolutionary search: a solution database that establishes diversity from the beginning, and then maintains it throughout the run; a smarter mutation router that plays into the strengths of large and small LLMs; and a rank-preserving proxy benchmark for rollout-heavy settings. Across systems-research benchmarks LEVI attains the highest score on a budget 3.3-6.7x smaller than the published frontier-model runs of existing frameworks like ShinkaEvolve, GEPA, and AdaEvolve; on one problem, LEVI matches the existing best at a 35x lower cost. On prompt optimization, LEVI matches or exceeds GEPA at less than half of its rollout budget on four different benchmarks. LEVI is available as an open-source framework at https://github.com/ttanv/levi.

---

### ConFit v3: Improving Resume-Job Matching with LLM-based Re-Ranking

Authors: Xiao Yu, Ruize Xu, Chengyuan Xue, Junyu Chen, Matthew So, Shijun Ma, Bo Liu, Xiangye Liang, Zhou Yu
Published: 2026-05-10
Categories: cs.CL
arXiv: https://arxiv.org/abs/2605.09760v1

A reliable resume-job matching system helps a company find suitable candidates from a pool of resumes and helps a job seeker find relevant jobs from a list of job posts. While recent advances in embedding-based methods such as ConFit and ConFit v2 can efficiently retrieve candidates at scale, the lack of controllability and explainability limits their real-world adaptations. LLM-based re-rankers can address these limitations through reasoning, but existing training recipes are developed on short-document benchmarks and do not account for noise in real-world recruiting data. In this work, we first conduct a systematic analysis over the LLM re-ranker training pipeline for person-job fit, covering inference algorithm design, RL algorithm selection, data processing, and SFT distillation. We find that using multi-pass re-ranking, training with listwise RL objectives, removing noisy samples, and distilling from a stronger LLM before RL significantly improves re-ranking performance. We then aggregate these findings to train ConFit v3 with Qwen3-8B and Qwen3-32B on real-world person-job fit datasets, and find significant improvements over existing best person-job fit systems as well as strong LLMs such as GPT-5 and Claude Opus-4.5. We hope our findings provide useful insights for future research on adapting LLM-based re-rankers to person-job fit systems.

---

### On Uniform Error Bounds for Kernel Regression under Non-Gaussian Noise

Authors: Johannes Teutsch, Oleksii Molodchyk, Marion Leibold, Timm Faulwasser, Armin Lederer
Published: 2026-05-10
Categories: cs.LG, stat.ML
arXiv: https://arxiv.org/abs/2605.09757v1

Providing non-conservative uncertainty quantification for function estimates derived from noisy observations remains a fundamental challenge in statistical machine learning, particularly for applications in safety-critical domains. In this work, we propose novel non-asymptotic probabilistic uniform error bounds for kernel-based regression. Compared to related bounds in the literature that are restricted to (conditionally) independent sub-Gaussian noise, our bounds allow to consider a broad class of non-Gaussian distributions, such as sub-Gaussian, bounded, sub-exponential, and variance/moment-bounded noise. Moreover, our results apply to correlated and uncorrelated noise. We compare our proposed error bounds with existing results in terms of the induced uncertainty region and their performance in safe control, demonstrating the tightness of the proposed bounds.

---

### Accelerating Power Method with Fast Sketching for Stronger Low-Rank Approximation

Authors: Shabarish Chenakkod, Michał Dereziński
Published: 2026-05-10
Categories: math.NA, cs.DS, cs.LG, stat.ML
arXiv: https://arxiv.org/abs/2605.09755v1

The power method is one of the most fundamental tools for extracting top principal components from data through low-rank matrix approximation. Yet, when the target rank is large, the cost of matrix multiplication associated with this procedure becomes a major bottleneck. We develop an algorithmic and theoretical framework for accelerating the power method using fast sketching, which is a popular paradigm in randomized linear algebra. Our framework leads to simple and provably efficient methods for singular value decomposition, low-rank factorization, and Nyström approximation, which attain strong numerical performance on benchmark problems. The key novelty in our analysis is the use of regularized spectral approximation, a property of fast sketching methods which proves more flexible in generalizing power method guarantees than traditional arguments.

---

### Learning from Acceptance: Cumulative Regret in the Game of Coding

Authors: Hanzaleh Akbari Nodehi, Parsa Moradi, Mohammad Ali Maddah-Ali
Published: 2026-05-10
Categories: cs.IT, cs.DC, cs.LG
arXiv: https://arxiv.org/abs/2605.09754v1

Classical coding-theoretic guarantees often rely on trust assumptions, such as requiring sufficiently many honest nodes compared with adversarial ones. These assumptions are difficult to enforce in open decentralized systems where participants are not centrally certified. At the same time, such environments often contain incentive mechanisms: participants may be rewarded only when their submitted data are accepted and the system remains functional. This changes the role of an adversary. Rather than acting as a pure saboteur, a strategic adversary may submit data that are consistent enough to be accepted while still degrading the quality of the final estimate. The game-of-coding framework models this strategic interaction between a data collector (DC) and an adversary. Existing works on the game of coding mostly consider the complete-information case, where the DC knows how the adversary trades off acceptance and estimation error. In this paper, we study an incomplete-information version of the game of coding in which the DC, acting as a Stackelberg leader, does not know the adversary's utility trade-off and must learn through repeated interaction. Prior work on the unknown-adversary setting considered an explore-then-commit objective, where only the final selected acceptance rule is evaluated. In contrast, we study the full learning trajectory: every acceptance rule used during the algorithm is executed and contributes to performance. We propose an algorithm that refines its search around promising acceptance rules, prove that it achieves sublinear cumulative regret, and evaluate its performance through numerical experiments.

---

### Language Models Without a Trainable Input Embedding Table: Learning from Fixed Minimal Binary Token Codes

Authors: A. Bochkov
Published: 2026-05-10
Categories: cs.CL
arXiv: https://arxiv.org/abs/2605.09751v1

Trainable input embedding tables are a standard component of modern language models. We ask whether they are actually necessary at the input interface. For a vocabulary of size $V$, exact token identity requires only $K=\lceil \log_2 V\rceil$ bits. We replace the usual trainable $V\times d_{\text{model}}$ input embedding matrix with fixed minimal binary token codes and a zero-parameter lift to model width. In our main setting, $V=65{,}536$, so $K=16$, and tokens are represented by fixed 16-dimensional binary codes tiled to $d_{\text{model}}=1024$. We also evaluate a fully table-free variant in which codes are generated from token IDs on the fly and randomly recoded by an invertible affine transform over $\mathbb{F}_2^K$. Across matched 32-layer decoder-only models trained on approximately 17B tokens and evaluated over three independent training seeds, fixed minimal codes achieve comparable held-out validation perplexity to a standard learned-input baseline while removing 67.1M trainable input parameters. The fixed-code runs have a lower mean validation perplexity in our experiments, 2.36 versus 2.44, but the observed gap is within the measured seed-to-seed variation of 4.8\%; we therefore interpret the result as evidence that the trainable input table is not necessary, rather than as a statistically resolved superiority claim. The table-free affine-recoded variant remains close at 2.39 despite a slightly shorter training run. These results show that, in this regime, a trainable input embedding table is not necessary for useful language modeling. The output projection remains standard and trainable.

---

### Fetal Brain Imaging: A Composite Neural Network Approach for Keyframe Detection in Ultrasound Videos

Authors: Aleksander Zamojski, Kacper Jarczak, Radoslaw Roszczyk
Published: 2026-05-10
Categories: cs.CV
arXiv: https://arxiv.org/abs/2605.09750v1

This article presents a novel approach to keyframe detection in ultrasound videos, with a particular focus on fetal brain imaging. The proposed model is a composite neural network architecture that combines a Convolutional Neural Network (CNN) with a Recurrent Neural Network (RNN). The CNN extracts spatial features from individual video frames, while the RNN captures temporal dependencies between consecutive frames within each video sequence. The proposed model may improve the efficiency and accuracy of fetal brain ultrasound analysis, thereby supporting earlier detection, diagnosis, and treatment planning for selected fetal brain conditions.

---

### Primal-Dual Guided Decoding for Constrained Discrete Diffusion

Authors: Federico Tomasi, Dmitrii Moor, Alice Wang, Mounia Lalmas
Published: 2026-05-10
Categories: cs.AI
arXiv: https://arxiv.org/abs/2605.09749v1

Discrete diffusion models generate structured sequences by progressively unmasking tokens, but enforcing global property constraints during generation remains an open challenge. We propose primal-dual guided decoding, an inference-time method that formulates constrained generation as a KL-regularised optimisation problem and solves it online via adaptive Lagrangian multipliers. At each denoising step, the method modifies token logits through an additive, constraint-dependent bias, with multipliers updated by mirror descent based on constraint violation. The bias arises as the optimal KL-regularised projection of the constraint, so the constrained distribution remains as close as possible to the model's unconstrained distribution while still satisfying the constraint. The method requires no retraining and no additional model evaluations beyond standard sampling, supports multiple simultaneous constraints, and provides formal bounds on constraint violation. We evaluate our approach on topical text generation, molecular design, and music playlist generation, showing that a single algorithm instantiated via domain-specific scoring functions improves constraint satisfaction while preserving relevant domain-specific quality metrics.

---

### Sequential Feature Selection for Efficient Landslide Segmentation from Multi-Spectral Data

Authors: Arsalaan Ahmad, Oktay Karakus, Paul L. Rosin
Published: 2026-05-10
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2605.09746v2

Landslide detection from satellite imagery has advanced through deep learning, yet most models rely on large, highly correlated spectral-topographic inputs whose contributions remain poorly understood. The question of which channels are actually necessary has received surprisingly little attention. This matters: redundant or correlated inputs obscure physical interpretability, inflate computational overhead, and can actively degrade model performance through the Hughes Phenomenon. We present a systematic, explainable channel-selection framework for the Landslide4Sense benchmark, combining Sentinel-2 multispectral and ALOS PALSAR terrain data with 16 engineered spectral and structural indices. Rather than relying on conventional single-band drop tests, which evaluate channels in isolation and miss interaction effects, we apply Sequential Forward Floating Selection (SFFS) to iteratively build and prune a candidate feature pool using a lightweight U-Net++ proxy model. Beyond identifying a compact 8-channel subset that matches or exceeds the segmentation F1 of configurations using up to 30 channels, we use the selection process itself to interrogate which spectral and topographic features landslide models genuinely rely on, and what this reveals about the physical cues driving their predictions. We argue that SFFS represents a principled feature selection approach to input design in Earth observation, in contrast to the prevailing practice of appending every available band and hoping the model learns what to ignore.

---

### Entropy-informed Decoding: Adaptive Information-Driven Branching

Authors: Benjamin Patrick Evans, Sumitra Ganesh, Leo Ardon
Published: 2026-05-10
Categories: cs.LG, cs.AI, cs.IT
arXiv: https://arxiv.org/abs/2605.09745v1

Large language models (LLMs) achieve remarkable generative performance, yet their output quality is dependent on the decoding strategy. While sampling-based methods (e.g., top-k, nucleus) and search-and-select based methods (e.g., beam search, best-of-n, majority voting) can improve upon greedy decoding, both approaches suffer from limitations: sampling generally commits to a single path, while search often expends excessive computation regardless of task complexity. To address these, we introduce Entropy-informed decoding (EDEN), a plug-and-play, model-agnostic decoding framework that adaptively allocates computation based on the model's own uncertainty, approximating higher-width beam search with fewer expansions. At each generation step, EDEN estimates the entropy of the output token distribution and adjusts the branching factor monotonically with the entropy, expanding more candidates in high-entropy regions and following a greedier path in low-entropy regions, improving token efficiency. Experiments across complex tasks, including mathematical reasoning, code generation, and scientific questions, demonstrate that EDEN consistently improves output quality over existing decoding strategies, achieving better accuracy-expansion trade-offs than fixed-width beam search. By treating next-token selection as a noisy maximisation problem, we prove that branching factors monotone in entropy are guaranteed to find better (i.e. more probable) continuations than any fixed branching factor within the same total expansion budget, and derive explicit regret rates characterising the benefit of the adaptive allocation.

---

### TIDES: Implicit Time-Awareness in Selective State Space Models

Authors: Taylan Soydan, Miguel A. Bessa, Dirk Mohr, Rui Barreira
Published: 2026-05-10
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2605.09742v1

Selective state space models (SSMs), such as Mamba, achieve strong per-token expressivity by making the time discretization step $\TildeΔ$ a learned function of the input. However, in doing so, $\TildeΔ$ ceases to represent a physical sampling interval, limiting its irregular time series modeling capability. Continuous-time SSMs, such as S5, preserve the physical meaning of $\TildeΔ$ and handle irregular timestamps natively ($\TildeΔ\equivΔ)$, but their dynamics remain linear time-invariant (LTI), limiting per-token expressivity. We propose \textbf{TIDES}, a selective SSM variant that reconciles selective and continuous architectures by moving input-dependence off the step size and onto the diagonal state matrix. As a result, $\TildeΔ$ retains its physical meaning, tied to the state discretization, allowing the model to handle irregular timestamps natively without sacrificing the per-token expressivity that makes selective SSMs effective. We show this on a novel \emph{Fading Flash} experimental benchmark, a compact controlled diagnostic for sequence models that jointly tests input-dependence and extrapolation to out-of-distribution $Δ$ values, and isolates the distinct failure modes of current state-of-the-art architectures that TIDES avoids by construction. On large-scale benchmarks, TIDES sets the new state-of-the-art average rank on UEA time-series classification and the Physiome-ODE regression benchmark. Code available at: https://github.com/TaylanSoydan/TIDES.

---

### LGB+: A Macroeconomic Forecasting Road Test

Authors: Philippe Goulet Coulombe
Published: 2026-05-10
Categories: econ.EM, stat.ME, stat.ML
arXiv: https://arxiv.org/abs/2605.09740v1

Needless to say, linear dynamics are pervasive in economic time series, particularly autoregressive ones. While gradient boosting with trees excels at capturing nonlinearities, it is inefficient in small samples when much of the predictive content is linear, expending splits to approximate relationships better captured by simple linear terms. This paper proposes LGB+, a boosting procedure operating on a more inclusive set of basis functions. The idea comes in two flavors. LGB+ evaluates a tree and a linear candidate at each step against out-of-bag data; only the winner advances. The simpler variant, LGB^A+, alternates on a fixed schedule: a block of tree updates, then a greedy linear correction, repeat. Both designs avoid ex ante commitments to any particular functional form or predictor selection. Because the prediction is the sum of a linear and a tree component, forecasts decompose natively into linear and nonlinear contributions, and so does permutation-based variable importance and historical proximity weights. In a quarterly U.S. macroeconomic forecasting exercise, LGB+ delivers strong gains for targets with pronounced autoregressive dynamics or mixed linear-nonlinear signals. Variables dominating the linear channel are those operating through autoregressive persistence or near-accounting relationships to the target (e.g., initial claims for unemployment and building permits for housing starts).

---

### The Silent Vote: Improving Zero-Shot LLM Reliability by Aggregating Semantic Neighborhoods

Authors: Sanket Badhe, Priyanka Tiwari, Deep Shah
Published: 2026-05-10
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2605.09739v1

Large Language Models are increasingly used as zero-shot classifiers in complex reasoning tasks. However, standard constrained decoding suffers from a phenomenon we define as Renormalization Bias. When a model is restricted to a small set of target labels, the standard softmax operation discards the probability mass assigned to semantic synonyms in the original distribution. This loss of information, which we call the Silent Vote, results in artificial overconfidence and poor calibration. We propose Semantic Softmax, an inference-time layer that recovers this lost information by aggregating the scores of the semantic neighborhood surrounding each target label. We evaluate this approach on Qwen-3 and Phi-4-mini models using GoEmotions and Civil Comments datasets. Our results demonstrate consistent improvements across all evaluation metrics: Semantic Softmax substantially reduces Expected Calibration Error (ECE) and Brier Score, while simultaneously enhancing discriminative performance in terms of AUROC and Macro-F1. By accounting for linguistic nuances, our method provides a more calibrated and accurate alternative for zero-shot classification.

---

### System-Prompt Anchoring with Cross-Attention Layers

Authors: Li Lixing
Published: 2026-05-10
Categories: cs.LG
arXiv: https://arxiv.org/abs/2605.09737v2

Cross-attention provides a dedicated route from a selected information source into a model's computation, but the effect of where that route is inserted remains underexplored. We study this question when the source is a privileged system-prompt span. We insert Cross-Attention Layer (CAL) blocks between the system prompt and text while keeping the causal-decoder backbone frozen. A ten-configuration sweep on a 1.5B backbone shows that performance is task-dependent and strongly affected by placement: later placements are generally more effective and parameter-efficient. In an 8B scaling study, we train only the overall best configuration and compare it with parameter-matched adaptation baselines. Across the evaluated benchmarks, the effects remain task-dependent: cross-attention changes instruction-following and security behavior while largely preserving general-task performance. Together, these experiments characterize placement as an important design variable when injecting system-prompt information through cross-attention.

---

### CALYREX: Cross-Attention LaYeR EXtended Transformers for System Prompt Anchoring

Authors: Li Lixing
Published: 2026-05-10
Categories: cs.LG
arXiv: https://arxiv.org/abs/2605.09737v1

Modern large language models (LLMs) rely on system prompts to establish behavioral constraints and safety rules. Standard causal self-attention treats privileged instructions and untrusted user content with equal structural priority -- a mismatch that leaves models vulnerable to prompt injection and instruction erosion over extended contexts. We propose CALYREX (Cross-Attention LaYeR EXtended transformers), which utilizes cross-attention between input and system prompt to structurally isolate and anchor the rule. A placement ablation on a 1.5B backbone identifies insertion at the final eighth of layers as optimal, confirmed by mechanistic activation analysis showing behavioral constraints are naturally concentrated there. At 8B scale, controlling for training data, backbone, and parameter budget, CALYREX yields $+7.4\%$ on instruction-following (IFEval) and $+16.3\%$ on multi-turn instruction adherence, while reducing many-shot jailbreaking attack success rate by $13\%$. This advantage appears to widen with model scale, consistent with larger models more effectively utilizing the dedicated routing pathway.

---

### KV-RM: Regularizing KV-Cache Movement for Static-Graph LLM Serving

Authors: Zhiqing Zhong, Zhijing Ye, Jian Zhang, Weijian Zheng, Bolun Sun, Xiaodong Yu
Published: 2026-05-10
Categories: cs.AR, cs.AI, cs.DC, cs.OS
arXiv: https://arxiv.org/abs/2605.09735v2

Static-graph LLM decoders provide predictable launches, fixed tensor shapes, and low submission overhead, but online decoding exposes highly irregular KV-cache behavior: request lengths differ, EOS events arrive asynchronously, and logical histories fragment over time. Dynamic runtimes recover flexibility through paged KV management and step-level scheduling, while static-graph executors often over-reserve memory and suffer burst-time latency outliers. This paper studies whether much of this variability can be absorbed below a fixed decode interface. We present KV-RM, a runtime design that regularizes KV-cache movement beneath a static-graph LLM decoder. KV-RM decouples logical KV histories from physical storage, tracks active KV state through a block pager, and materializes each decode step through a single committed descriptor. A merge-staged transport path coalesces non-contiguous KV mappings into a small number of large transfer groups before a fixed-shape attention kernel consumes them. Optional bounded far-history summaries can be enabled under the same interface, but the core design does not depend on them. On a 2-GPU NVIDIA A100 node, KV-RM improves mixed-length decoding throughput and tail latency relative to a static-graph baseline, reduces reserved KV memory across workload families, and removes severe burst-time latency spikes under production-trace replay. These results suggest that KV-cache movement, rather than kernel shape, can be an effective boundary for recovering runtime flexibility in static-graph LLM serving.

---

### Trajectory Supervision for Continual Tool-Use Learning in LLMs

Authors: Vishnu Vardhan Reddy, Sagnik Chatterjee, Soumik Bhatta
Published: 2026-05-10
Categories: cs.SE, cs.AI, cs.MA
arXiv: https://arxiv.org/abs/2605.09734v1

Most language-model training data shows final artifacts, not the process that produced them. We study a tractable version of this question in tool use: when a model learns a stream of new API domains, does keeping tool-use trajectories help compared with stripping the intermediate API trace? We fine-tune Llama 3.1 8B Instruct with QLoRA on API-Bank using four sequential domain blocks. Condition A strips previous API request/response lines from the prompt and trains the model to predict the next API call. Condition B keeps the trajectory context. In a single-seed pilot, full held-out generation evaluation shows that Condition B reaches 56.9\% final exact full-call accuracy compared with 39.2\% for Condition A. B also improves final API-name accuracy by 7.7 points. However, B uses 25.1\% more training tokens, the run uses one seed, and the task is next-call prediction rather than full dialogue success.

---

### RubricRefine: Improving Tool-Use Agent Reliability with Training-Free Pre-Execution Refinement

Authors: Will LeVine, Brendan Evers, Sam Saltwick, Abhay Venkatesh
Published: 2026-05-10
Categories: cs.LG, cs.SE
arXiv: https://arxiv.org/abs/2605.09730v5

Iterative self-refinement is a popular inference-time reliability technique, but its effectiveness in code-mode tool use depends heavily on the structure of the feedback signal: unstructured critique helps inconsistently across models, and even revision with real execution feedback improves only modestly. The dominant failures are inter-tool contract violations (wrong output shape, incorrect tool routing, broken argument provenance) that run to completion without raising errors, making runtime feedback insufficient. We introduce RubricRefine, a training-free method for pre-execution contract checking that generates task- and registry-specific rubrics, scores candidate code against explicit contract checks, and iteratively repairs failures before any execution occurs. RubricRefine reaches $0.86$, averaged across seven models, on M3ToolEval with zero execution attempts, improving over prior inference-time baselines at lower latency than rubric-guided reranking. Performance remains flat on the predominantly single-step API-Bank, consistent with the method's reliance on inter-tool contract structure. Results on AppWorld demonstrate that our method maintains an advantage in the multi-turn setting. Because the rubric is derived from the supplied tool documentation, the method's advantage survives incomplete documentation but reverses under incorrect documentation. A rubric-category ablation identifies which rules are load-bearing, and top-bin calibration enables early stopping even where aggregate calibration is poor.

---

### RubricRefine: Improving Tool-Use Agent Reliability with Training-Free Pre-Execution Refinement

Authors: Will LeVine, Brendan Evers, Sam Saltwick, Abhay Venkatesh
Published: 2026-05-10
Categories: cs.LG, cs.SE
arXiv: https://arxiv.org/abs/2605.09730v3

Iterative self-refinement is a popular inference-time reliability technique, but its effectiveness in code-mode tool use depends heavily on the structure of the feedback signal: unstructured critique helps inconsistently across models, and even revision with real execution feedback improves only modestly ($0.75$ vs. $0.65$ baseline). The dominant failures are inter-tool contract violations (wrong output shape, incorrect tool routing, broken argument provenance) that run to completion without raising errors, making runtime feedback insufficient. We introduce RubricRefine, a training-free method for pre-execution semantic contract verification that generates task- and registry-specific rubrics, scores candidate code against explicit contract checks, and iteratively repairs failures before any execution occurs. RubricRefine reaches $0.86$, averaged across seven models, on M3ToolEval with zero execution attempts, improving over prior inference-time baselines with up to $2.6\times$ lower latency. Performance remains flat on the predominantly single-step API-Bank, consistent with the method's reliance on inter-tool contract structure. A rubric-category ablation and calibration analysis further characterize when and why the method works.

---

### One for All: A Non-Linear Transformer can Enable Cross-Domain Generalization for In-Context Reinforcement Learning

Authors: Bowen He, Juncheng Dong, Lin Lin, Xiang Cheng
Published: 2026-05-10
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2605.09727v1

A central challenge in reinforcement learning (RL) is to learn models that generalize beyond the tasks on which they are trained, a goal traditionally pursued through multi-task and meta RL. Recently, transformer architectures have emerged as a promising approach, enabling adaptation to new tasks via in-context learning without explicit parameter updates. From a functional perspective, a transformer can be viewed as a functional operator that maps a context to a task-specific function. It is thus fundamental to understand and design this operator to support stronger generalization in RL. In this work, we address this resulting question of generalization from a kernel-based perspective by establishing a connection between non-linear transformers and kernel-based temporal difference learning. By interpreting the transformer as performing regression in a Reproducing Kernel Hilbert Space (RKHS), we show that value functions from different domains can be represented using a shared set of weights, provided they lie within the same RKHS. Experiments on multiple MetaWorld domains support this interpretation, demonstrating convergence of the temporal-difference objective.

---

### On-Policy Distillation with Best-of-N Teacher Rollout Selection

Authors: Ke Zhang, Yunjie Tian, Dongdi Zhao, Yijiang Li, Yuanye Liu, Vishal M Patel, Di Fu
Published: 2026-05-10
Categories: cs.CV
arXiv: https://arxiv.org/abs/2605.09725v3

On-policy distillation (OPD), which supervises a student on its own sampled trajectories, has emerged as a data-efficient post-training method for improving reasoning while avoiding the reward dependence of reinforcement learning and the catastrophic forgetting often observed in standard supervised fine-tuning. However, standard OPD typically computes teacher supervision under noisy student-generated contexts and often relies on a single stochastic teacher rollout per prompt. As a result, the supervision signal can be high-variance: the sampled teacher trajectory can be incorrect, uninformative, or poorly matched to the student's current reasoning behavior. To address this limitation, we propose BRTS, a Best-of-N Rollout Teacher Selection framework for on-policy distillation. BRTS augments standard student-context OPD with a teacher-context supervision branch constructed from the curated teacher trajectory. Rather than distilling from the first sampled teacher rollout, BRTS samples a small pool of teacher trajectories and selects the auxiliary trajectory using a simple priority rule: correctness first, student alignment second. When multiple correct teacher trajectories are available, BRTS chooses the one most aligned with the student's current behavior; when unconditioned teacher samples fail on harder prompts, it invokes a ground-truth-conditioned recovery step to elicit a natural derivation. The selected trajectory is then used to provide reliable teacher-context supervision inside the OPD loop, augmented with an auxiliary loss on the teacher trajectory. Experiments on AIME 2024, AIME 2025, and AMC 2023 show that BRTS improves over standard OPD on challenging reasoning benchmarks, with the largest gains on harder datasets. Our code is available at https://github.com/BWGZK-keke/BRTS.

---

### On-Policy Distillation with Best-of-N Teacher Rollout Selection

Authors: Ke Zhang, Yunjie Tian, Dongdi Zhao, Yijiang Li, Yuanye Liu, Vishal M Patel, Di Fu
Published: 2026-05-10
Categories: cs.CV
arXiv: https://arxiv.org/abs/2605.09725v2

On-policy distillation (OPD), which supervises a student on its own sampled trajectories, has emerged as a data-efficient post-training method for improving reasoning while avoiding the reward dependence of reinforcement learning and the catastrophic forgetting often observed in standard supervised fine-tuning. However, standard OPD typically computes teacher supervision under noisy student-generated contexts and often relies on a single stochastic teacher rollout per prompt. As a result, the supervision signal can be high-variance: the sampled teacher trajectory can be incorrect, uninformative, or poorly matched to the student's current reasoning behavior. To address this limitation, we propose BRTS, a Best-of-N Rollout Teacher Selection framework for on-policy distillation. BRTS augments standard student-context OPD with a teacher-context supervision branch constructed from the curated teacher trajectory. Rather than distilling from the first sampled teacher rollout, BRTS samples a small pool of teacher trajectories and selects the auxiliary trajectory using a simple priority rule: correctness first, student alignment second. When multiple correct teacher trajectories are available, BRTS chooses the one most aligned with the student's current behavior; when unconditioned teacher samples fail on harder prompts, it invokes a ground-truth-conditioned recovery step to elicit a natural derivation. The selected trajectory is then used to provide reliable teacher-context supervision inside the OPD loop, augmented with an auxiliary loss on the teacher trajectory. Experiments on AIME 2024, AIME 2025, and AMC 2023 show that BRTS improves over standard OPD on challenging reasoning benchmarks, with the largest gains on harder datasets. Our code is available at https://github.com/BWGZK-keke/BRTS.

---

### Model Capacity Determines Grokking through Competing Memorisation and Generalisation Speeds

Authors: Yiding Song, Hanming Ye
Published: 2026-05-10
Categories: cs.LG
arXiv: https://arxiv.org/abs/2605.09724v1

Existing accounts of grokking explain the phenomena in terms of mechanistic frameworks such as circuit efficiency or lazy-to-rich transitions. However, despite a known dependence between grokking and model size, how model capacity shapes grokking remains an open question. We give an information-theoretic account of this relationship on the task of modular arithmetic, showing that grokking does not immediately occur when a model becomes large enough to memorise the training set, but rather emerges as the outcome of a competition between two measurable timescales: a memorisation speed $T_{\text{mem}}(P)$ and a generalisation speed $T_{\text{gen}}(P)$, both of which are functions of model parameter count $P$. Adapting the information capacity framework of Morris et al. (2025), we estimate $T_{\text{mem}}(P)$ on random-label data of equivalent complexity and $T_{\text{gen}}(P)$ on the modular task itself, and show that grokking emerges close to the parameter scale where these timescales intersect. The framework also suggests an empirical model for predicting memorisation speed given model capacity and dataset complexity, recovering the previously reported empirical observation that larger models memorise faster. Overall, we motivate the formalisation of different learning timescales as important abstractions to study when explaining how model capacity shapes grokking on algorithmic tasks.

---

### Benchmarking Transformer and xLSTM for Time-Series Forecasting of Heat Consumption

Authors: Marja Wahl, Daniel R. Bayer, Sven Rausch, Marco Pruckner
Published: 2026-05-10
Categories: cs.LG
arXiv: https://arxiv.org/abs/2605.09722v1

Obtaining an accurate short-term forecasting for heat demand is an essential part of operating district heating networks cost-efficient and reliable. Heat consumption time series at the building level are highly dependent on exogenous variables such as outdoor temperature and individual usage patterns, making forecasting in this context a challenging task. Thus, this paper benchmarks novel Transformer-based and xLSTM architectures for short-term heat-demand forecasting. Using hourly data from 25 German buildings (2017-2025), we compare three-hour and 24-hour forecasting horizons relevant for intraday control and day-ahead scheduling. We establish a multi-building benchmark that tests whether models trained on pooled, heterogeneous building data are able to generalize across diverse building stock. The results show that the xLSTM achieves the lowest RMSE (19.88 kWh for three-hour, 21.47 kWh for 24-hour forecasts), while the Temporal Fusion Transformer attains the best MAE (9.16 kWh for three-hour forecasts). As xLSTMs and Transformers require long training times and have a huge number of trainable parameters, their sustainability remains questionable. Therefore, this paper further investigates the trade-off between predictive accuracy and computational resource demand of the evaluated forecasting models. The findings indicate that also low-parameter models like a traditional fully-connected network achieve good predictive results, highlighting that marginal accuracy gains of the novel prediction models come at substantial resource expense for this use case.

---

### Security Risks in Tool-Enabled AI Agents: A Systematic Analysis of Privileged Execution Environments

Authors: Hardik Goel
Published: 2026-05-10
Categories: cs.CR, cs.AI
arXiv: https://arxiv.org/abs/2605.09721v1

Tool-enabled AI agents are increasingly deployed in cloud-hosted environments and offered as services, where they perform side-effecting operations through privileged tools within execution environments. While such agents enable powerful automation, the security implications of hosting autonomous agents in privileged execution environments are not yet fully explored. This paper presents a structured analysis of security risks associated with cloud-hosted AI agents. We introduce a taxonomy of risk categories, illustrate these risks through three representative agent scenarios, and discuss mitigation strategies along with their tradeoffs. A small controlled experiment empirically illustrates risk manifestation and the effect of lightweight mitigations in this setup. Our analysis suggests that many risks in autonomous cloud agents arise not from novel vulnerabilities, but from over-privileged tools, capability-intent mismatches, and ambient authority leakage in execution environments. Based on these findings, we derive practical design guidelines for deploying AI agents in the cloud more securely.

---

### Distilling 3D Spatial Reasoning into a Lightweight Vision-Language Model with CoT

Authors: Alaa Asfour, Christopher Indris, Leihan Chen, Tejas Vyas, Guanghui Wang
Published: 2026-05-10
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2605.09719v1

Large-scale 3D vision-language models (VLMs) like LLaVA-3D offer strong spatial reasoning but are difficult to deploy due to high computational costs. We propose a knowledge distillation framework that transfers spatial reasoning from a 7B teacher to a 2.29B student model. Our approach achieves 8.7x lower inference latency and a 3x reduction in model size while retaining 54-72% of the teacher's performance. The framework utilizes VGGT as the vision encoder and a multi-task distillation pipeline with uncertainty-aware loss weighting. To improve reasoning without chain-of-thought (CoT) data, we introduce "Hidden CoT": learnable latent tokens that serve as an internal scratchpad before answer generation. This is the first use of latent scratchpad reasoning in distilled 3D VLMs. The student model jointly performs spatial description, depth estimation, and object detection. Experiments on ScanNet and 3D-FRONT show strong spatial understanding, reaching 68-72% accuracy on proximity and contact tasks. Our framework enables efficient 3D scene QA on resource-constrained platforms.

---

### Learning stochastic multiscale models through normalizing flows

Authors: Anan Saha, Arnab Ganguly
Published: 2026-05-10
Categories: stat.ML, cs.LG, math.PR, math.ST
arXiv: https://arxiv.org/abs/2605.09718v1

Many systems in physics, engineering, and biology exhibit multiscale stochastic dynamics, where low-dimensional slow variables evolve under the influence of high-dimensional fast processes. In practice, observations are often limited to a single trajectory of the slow component, while the fast dynamics remain unobserved, making statistical learning challenging. Approaches based on partial differential equations (PDE), such as Fokker-Planck formulations, aim to characterize the evolution of probability densities, typically requiring dense space-time data or grid-based solvers. In contrast, we adopt a trajectory-based perspective and develop a data-driven framework for learning effective stochastic dynamics from a single observed path. We model the dynamics by coupled multiscale stochastic differential equations (SDEs) and first obtain a principled model reduction through stochastic averaging. Unlike generic model reduction techniques such as PCA, this respects the dynamical structure of the original system and explicitly incorporates the interaction between slow and fast scales. A central challenge, however, is that the reduced model depends on the invariant distribution of the fast process, which is a solution to an intractable and often unknown PDE. We introduce a novel learning framework that parameterizes the invariant distribution using normalizing flows, enabling expressive density modeling in the latent fast-variable space. The flow is trained end-to-end by optimizing a penalized likelihood objective induced by the reduced stochastic dynamics. Furthermore, we develop a Bayesian variational inference procedure for uncertainty quantification, employing a second normalizing flow to approximate the posterior distribution over model parameters. This yields a scalable approach to capturing epistemic uncertainty in multiscale systems.

---

### Medical Model Synthesis Architectures: A Case Study

Authors: Katherine M. Collins, Marlene Berke, Ilia Sucholutsky, Ayman Ali, Adrian Weller, Timothy J. O'Donnell, Tyler Brooke-Wilson, Lionel Wong, Joshua B. Tenenbaum
Published: 2026-05-10
Categories: cs.AI
arXiv: https://arxiv.org/abs/2605.09716v1

Medicine is rife with high-stakes uncertainty. Doctors routinely make clinical judgments and decisions that juggle many fundamental unknowns, like predictions about what might be causing a patients' symptoms or decisions about what treatment to try next. Despite increasing interest in developing AI systems that aid or even replace doctors in clinical settings, current systems struggle with calibrated reasoning under uncertainty, and are often deeply opaque about their reasoning. We propose a framework for AI systems that can make practically useful but formally transparent clinical predictions under uncertainty. Given a clinical situation, our framework (MedMSA) uses language models to retrieve relevant prior knowledge, but constructs a formal probabilistic model to support calibrated and verifiable inferences under uncertainty. We show how an initial proof-of-concept of this framework can be used for differential diagnosis, producing an uncertainty-weighted list of potential diagnoses that could explain a patients' symptoms, and discuss future applications and directions for applying this framework more generally for safe clinical collaborations.

---

### Quantifying the Risk-Return Tradeoff in Forecasting

Authors: Philippe Goulet Coulombe
Published: 2026-05-10
Categories: econ.EM, q-fin.PM, stat.ML
arXiv: https://arxiv.org/abs/2605.09712v1

Average forecast accuracy is not the same as forecast reliability. I treat forecast loss differentials relative to a benchmark as a return series. I then evaluate these returns using risk-adjusted performance measures from finance, including the Sharpe ratio, Sortino ratio, Omega ratio, and drawdown-based metrics. I also introduce the Edge Ratio capturing a model's propensity to deliver uniquely informative predictions relative to the forecasting frontier. I apply this framework to U.S. macroeconomic forecasting, comparing econometric benchmarks, machine learning models, a foundation model (TabPFN), and the Survey of Professional Forecasters. While it is often feasible to beat professional forecasters in terms of average accuracy, it is much harder to beat them on a risk-adjusted basis. They rarely exhibit catastrophic failures and often achieve high Edge Ratios, plausibly reflecting the value of contextual judgment. Nonetheless, selected machine learning methods deliver attractive risk profiles for specific targets. The framework naturally extends to meta-analyses across targets, horizons, and samples, illustrated with a density forecast evaluation and the M4 competition.

---

### Metal-Sci: A Scientific Compute Benchmark for Evolutionary LLM Kernel Search on Apple Silicon

Authors: Víctor Gallego
Published: 2026-05-10
Categories: cs.LG, cs.AI, cs.DC
arXiv: https://arxiv.org/abs/2605.09708v2

We present Metal-Sci, a 10-task benchmark of scientific Apple Silicon Metal compute kernels spanning six optimization regimes (stencils, all-pairs in $n$-body problems, multi-field Boltzmann, neighbor-list molecular dynamics, multi-kernel PDE, FFT). Each task ships a CPU reference, a roofline-anchored fitness function, and a held-out generalization size. We pair the benchmark with a lightweight harness for automatic kernel search that runtime-compiles each candidate, scores it against the roofline across multiple sizes, and feeds structured compile and per-size correctness diagnostics back to a frozen LLM driving a $(1{+}1)$ evolutionary loop. We report matched single-model sweeps of Claude Opus 4.7, Gemini 3.1 Pro, and GPT 5.5 on M1 Pro: in-distribution self-speedups span $1.00\times$ to $10.7\times$. Beyond raw speedup, our central methodological claim is structural: the held-out gate scoring function $Φ_\mathcal{T}$ (evaluated once at end-of-run on a configuration the agent never sees during search) functions as a cheap mechanical oversight primitive on this automatic search loop, catching e.g. an Opus template <uint D> HMC win that returns wrong samples at unseen dimensions, and a GPT FFT3D best that wins in-distribution at $2.95\times$ speedup but collapses to $0.23\times$ on a $256^3$ held-out cube, a silent regression that the in-distribution score alone cannot see. Code at https://github.com/vicgalle/metal-sci-kernels

---

### Adaptive Data Harvesting for Efficient Neural Network Learning with Universal Constraints

Authors: Siteng Kang, Xinhua Zhang
Published: 2026-05-10
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2605.09707v1

Training neural networks to satisfy universal constraints over continuous domains poses unique challenges. Common examples include Lyapunov Neural Networks (Lyapunov NNs) and Physics-Informed Neural Networks (PINNs), where analytical solutions are generally either unavailable or overly restrictive. Sample-based methods are therefore commonly used to enforce these constraints, and the choice of samples has a substantial impact on convergence speed, stability, and solution quality. Most existing methods rely on fixed heuristics or handcrafted rules, and are suboptimal in practice. In this paper, we aim to improve upon them by learning, from data and experience, how to dynamically and iteratively adjust the samples in response to the model's evolving learning performance. Trained by reinforcement learning, the learned policy improves empirical constraint satisfaction on test problems while significantly improving efficiency. We validate the approach on both Lyapunov NNs and PINNs, and demonstrate its broader applicability to domains where adaptive input selection is essential for effective training.

---

### MOTOR-Bench: A Real-world Dataset and Multi-agent Framework for Zero-shot Human Mental State Understanding

Authors: Xiaoyu Yuan, Niklas Heikkala, Tiina Törmänen, Hanna Järvenoja, Guoying Zhao, Haoyu Chen
Published: 2026-05-10
Categories: cs.CV
arXiv: https://arxiv.org/abs/2605.09703v1

Understanding human mental states from natural behavior is crucial for intelligent systems in the real world. However, most current research focuses on predicting isolated mental state labels, lacking structured annotations of complex interpersonal interactions. To support structured analysis, we introduce MOTOR-Bench, a carefully-designed benchmark with a real-world dataset MOTOR-dataset, containing 1,440 multimodal video clips in collaborative learning scenarios, reflecting key real-world data challenges including natural class imbalance, visual noise, and domain-specific language. Each sample is labeled by educational experts based on self-regulated learning theory. We further evaluate several state-of-the-art multimodal large language models and multi-agent systems in a zero-shot setting on our MOTOR-Bench. However, their performance on this task remains limited, suggesting that existing methods still struggle with structured reasoning from observable behavior to deeper mental states. To address this challenge, we propose a reasoning multi-agent framework, named MOTOR-MAS. It coordinates multiple agents through a structured agent coordination mechanism to infer explicit behaviors, internal cognitions, and psychological emotions. Experimental results show that our MOTOR-MAS outperforms the best single-model benchmark by 15.93 points in Macro-F1 scores for the three labels of behavior, cognition, and emotion, and outperforms the general multi-agent benchmark by 10.2 points in internal cognition prediction.

---

### Calibrate, Don't Curate: Label-Efficient Estimation from Noisy LLM Judges

Authors: Yanran Li
Published: 2026-05-10
Categories: stat.ME, cs.CL
arXiv: https://arxiv.org/abs/2605.09702v1

Multi-judge evaluation is increasingly used to assess LLMs and reward models, and the prevailing heuristic is to curate: keep the most accurate judges and discard weaker ones. We show that this heuristic can reverse when the target is not point accuracy, but calibrated probabilistic evaluation from a labeled calibration set. Holding the aggregation and calibration procedures fixed, we compare accuracy-ranked top-$k$ judge selection with using the full judge panel. Across four labeled pairwise-evaluation benchmarks spanning LLM-as-judge and reward-model settings, the calibrated full panel consistently outperforms accuracy-based selection. On RewardBench2, retaining all judges achieves negative log-likelihood (NLL) of $0.006$ versus $0.013$ under top-5 selection, halving the calibration error. This advantage persists after judge-family deduplication and against stronger same-pipeline subset search. We explain this reversal with oracle analyses showing that the optimal calibrated risk under proper scoring rules cannot increase when additional judge signals are made available, and that even below-chance judges can be useful when their biases are learnable and their signals are non-redundant. The resulting operating principle is simple: in multi-judge evaluation with labeled calibration data, do not discard weak judges by accuracy alone; keep them when they are parseable, non-redundant, and calibratable.

---