中 EN
Sign In Register

Academic Intelligence · Curated Daily

Explore the Frontier of Global Academia

AcademicHub aggregates real-time literature from top journals and preprint platforms. Build your personal research radar and let large language models compile cross-disciplinary analysis briefings automatically.

Sign In Register
01.
arXiv (CS.CL) 2026-06-18

Enhancing Multilingual Reasoning via Steerable Model Merging

Authors:
Zhuoran Li↗Rui Xu↗Jian Yang↗Junnan Liu↗Zhijun Chen↗Qianren Mao↗Hongcheng Guo↗Jiaheng Liu↗Likang Xiao↗Ming Li↗Xiaojie Wang↗

Model merging is an effective technique for composing the capabilities of a multilingual model and a reasoning model. It has achieved promising generalization in multilingual reasoning tasks by aligning feature spaces of different models. However, the merged single model often fails to address the conflicts between source models, leading to suboptimal performance. In other words, the one-size-fits-all merging strategy may not align with the characteristics of different inputs which may require prioritizing certain models over others. To this end, we propose a Steerable Model Merging (ST-Merge) framework to modulate the contribution of each source model. To realize this idea, we introduce a gated cross-attention mechanism to weight or filter the two attended source models in an adaptive manner. Extensive experiments demonstrate that ST-Merge consistently outperforms multiple strong baselines on four multilingual reasoning benchmarks across 21 different languages.

Read & Discuss → View Source →
02.
arXiv (CS.AI) 2026-06-15

Learning optimal policies from event logs through reinforcement learning: a comparison of deep and MDP-based approaches

Authors:
Stefano Branchi↗Andrei Buliga↗Chiara Di Francescomarino↗Chiara Ghidini↗Riccardo Graziosi↗Francesca Meneghello↗Massimiliano Ronzani↗

arXiv:2303.09209v2 Announce Type: replace Abstract: Prescriptive Process Monitoring is an emerging area within Process Mining that focuses on recommending actions to optimize business outcomes. Most existing works prescribe pre-defined interventions, i.e., sets of actions applied to ongoing process executions to achieve a specific objective or Key Performance Indicator (KPI). In contrast, only a few approaches have explored learning and evaluating optimal behavioral policies, i.e., general strategies that determine the best sequence of actions to maximize a desired KPI. In this paper, we address the problem of learning optimal behavioral policies by proposing an AI-based approach that learns an optimal policy directly from historical process executions using Reinforcement Learning (RL) to recommend the best actions for optimizing a KPI. To this end, we employ two RL techniques. The first is a classical model-based approach that extends previous work by the authors through the construction of a Markov Decision Process (MDP) capturing process behavior. The second is a model-free technique based on offline Deep RL. Unlike state-of-the-art work, we aim to minimize the use of domain knowledge and learn optimal policies directly from historical event data. This allows us to learn when to apply interventions and discover effective ones directly from data. Moreover, we target complex scenarios involving external actors, where the process owner controls only part of the activities. We adopt a data-driven Business Process Simulation (BPS) environment to evaluate the learned policies. Results show that both methods improve the targeted KPI with similar effectiveness, while the model-based approach outperforms offline Deep RL in computational efficiency.

Read & Discuss → View Source →
03.
arXiv (CS.CL) 2026-06-12

SkMTEB: Slovak Massive Text Embedding Benchmark and Model Adaptation

Authors:
Marek \v{S}uppa↗Andrej Ridzik↗Daniel Hl\'adek↗Nat\'alia K\v{n}a\v{z}ekov\'a↗Vikt\'oria Ondrejov\'a↗

We introduce SkMTEB, the first comprehensive MTEB-style text embedding benchmark for Slovak, a low-resource West Slavic language, comprising 31 datasets across 7 task types – nearly 4$\times$ the depth of existing multilingual benchmark coverage for Slovak. Our evaluation of 31 embedding models reveals that large instruction-tuned multilingual models achieve the strongest performance, while existing Slovak-specific models trained for NLU tasks transfer poorly to embedding tasks. To address the need for efficient, locally-deployable Slovak embeddings, we develop \texttt{e5-sk-small} (45M parameters) and \texttt{e5-sk-large} (365M) by applying vocabulary trimming and fine-tuning to Multilingual E5 models. Despite size reductions of up to 62\%, our open-source models achieve competitive performance with proprietary APIs while remaining locally deployable for semantic search and retrieval-augmented generation (RAG). We release the benchmark, models, datasets, and code openly, hoping our approach offers a replicable path for other under-resourced languages.

Read & Discuss → View Source →
04.
bioRxiv (Bioinfo) 2026-06-18

Metrics for Evaluating Biological AI Model Predictive Accuracy at the Data-Substrate Level

Authors:
Ewing↗M. A↗

Reports in the biological literature disagree on whether a given model can predict a biological outcome from a given data sample — one study finding a model capable, another, on the same kind of data, finding it is not. This is particularly a challenge in relation to LLMs–where the models are large and opaque, with weights and training data inaccessible.textbf{ }Such disagreements cannot be settled by directly inspecting the model. To address this challenge, we considertextbf{ }an alternative approach: assessing whether the data sample is adequate to support the prediction asserted. For a given dataset, its substrate — the underlying structure of the data — determines what any model can recover, independent of architecture or capacity. At the same time, predicting the present state of a biological process and predicting the direction of its future change are different tasks; the second is supportable among AI models only where the data encode direction as determinable from the state — a property we call encoding — and is unsupportable where the same observed state precedes change in opposite directions — a property we call non-identifiability, in the informational rather than the statistical sense. We introduce two generic metrics, Predictive Blindness Risk (PBR) and Prediction Indeterminacy Measure (PIM), that evaluate a data substrate for predictive accuracy directly — without access to model weights, architecture, or training data — and locate the regions of a data substrate where a predictive claim can be supported and where it cannot. Using human biological subjects, we employ the Yale Brain Metastases Longitudinal Data (1,430 human subjects; 11,892 MRI studies; four sequences) and show that direction of change was non-identifiable across regions encompassing the majority of transitions; a nonlinear AI model gained essentially nothing over majority-direction prediction there while recovering direction near-perfectly where the state encoded it; and model accuracy tracked data-substrate resolvability continuously (Spearman {rho} = -0.95 to -1.00). The metrics adjudicate, before any model is trusted and from the data alone, where claims of predictive accuracy — of state, or of the law of change — can be supported.

Read & Discuss → View Source →
05.
arXiv (CS.CL) 2026-06-11

Substrate Asymmetry in User-Side Memory: A Diagnostic Framework

Authors:
Youwang Deng↗

User-side memory in LLMs is typically scored as a single "personalization" capability: given a user's history, is the output more user-aware? We show this aggregate metric hides opposite-direction failures. Memory factorises into at least three orthogonal axes – behavioral consistency (style, voice), factual presence (recall facts in history), and factual absence (abstain when a fact is absent) – and no single substrate wins all three. Comparing per-user gamma-LoRA (a small LoRA adapter trained on each user's history; gamma denotes per-user, not per-task) against BGE-large dense top-K retrieval on a controlled 50-user synthetic corpus and a real-data probe (LaMP-3), we find gamma-LoRA decisively wins behavioral style while RAG decisively wins factual absence – and the same query-projection cells in attention layers 21-35 causally load-bear both effects in opposite directions (zeroing those LoRA weights raises absence-probe TPR by +33 pp and drops presence-probe TPR by 20 pp). On the more heavily RLHF-tuned Llama-3.1-8B-Instruct the asymmetry strengthens, not heals: parametric memory's behavioral advantage collapses while its absence-calibration deficit against retrieval widens – an alignment tax on parametric user-memory. On real-data LaMP-3, gamma-LoRA underperforms a majority baseline; a 9-condition mitigation sweep diagnoses this as instruction-following collapse, not substrate failure (a 9x2 cross-product shows the eval-time {1..5} logit mask drives main_acc to >=0.995 on every recipe), and the best training-time fix replicates bit-identically on Llama. Finally, substrate-selection routing is question-classification, not calibration: a 110M DistilBERT on the question text alone beats every logit-based router. We contribute the diagnostic framework, the diagnosed real-data negative, the alignment-tax replication, and the routing-as-classification finding.

Read & Discuss → View Source →
06.
arXiv (CS.LG) 2026-06-17

Stable and Steerable Sparse Autoencoders with Weight Regularization

Authors:
Piotr Jedryszek↗Oliver M. Crook↗

arXiv:2603.04198v2 Announce Type: replace-cross Abstract: Sparse autoencoders (SAEs) are widely used to extract human-interpretable features from neural network activations, but their learned features can vary substantially across random seeds and training choices. To improve stability, we studied weight regularization by adding L1 or L2 penalties on encoder and decoder weights, and evaluate how regularization interacts with common SAE training defaults. On MNIST, we observe that L2 weight regularization produces a core of highly aligned features and, when combined with tied initialization and unit-norm decoder constraints, it dramatically increases cross-seed feature consistency. For TopK SAEs trained on language model activations (Pythia-70M-deduped), adding a small L2 weight penalty increased the fraction of features shared across three random seeds and roughly doubles steering success rates, while leaving the mean of automated interpretability scores essentially unchanged. Finally, in the regularized setting, activation steering success becomes better predicted by auto-interpretability scores, suggesting that regularization can align text-based feature explanations with functional controllability.

Read & Discuss → View Source →
07.
arXiv (CS.CV) 2026-06-12

Towards More General Control of Diffusion Models Using Jeffrey Guidance

Authors:
Rapha\"el Razafindralambo↗R\'emy Sun↗Fr\'ed\'eric Precioso↗Jes Frellsen↗Pierre-Alexandre Mattei↗

A key strength of diffusion models lies in their flexibility, since their outputs can be controlled at sampling time through guidance. However, beyond simple cases such as conditional sampling, the target distribution is often left implicit, defined only through a sampling rule or a heuristic energy function. To address this, we propose Jeffrey guidance, a principled framework that extends diffusion-model control to applications beyond what standard guidance can express. It leverages Jeffrey's rule of conditioning to update marginal distributions towards a prescribed target, preserving the conditional structure and minimally perturbing the joint distribution. We first demonstrate Jeffrey guidance by targeting a prescribed embedding distribution. With Inception embeddings as the target, this leads to substantial reductions in FID on both CIFAR-10 and FFHQ. We further apply Jeffrey guidance to fairness on CelebA-HQ, updating an unconditional diffusion model to enforce independence between attributes.

Read & Discuss → View Source →
08.
medRxiv (Medicine) 2026-06-19

Reassessing Instrument Strength in Two-Sample Mendelian Randomization Analysis

Authors:
Liu↗Huang↗Y.-J↗Purushotham↗Sofer↗

Mendelian randomization (MR) analysis is widely used to estimate causal relationships between risk factors and outcomes of interest. Two-sample MR approaches have gained increasing attention in genetic epidemiology due to the growing availability of Genome-Wide Association Study (GWAS) summary statistics from public databases. A critical step in two-sample MR is the selection of genetic variants as instrumental variables (IVs). Although genome-wide significant variants are typically preferred, the inclusion of variants with weaker association p-values is considered, as they may potentially improve power through an increased instrument number of instruments, while they may introduce weak instrument bias and attenuate effect estimates towards the null. Our simulation results show that even modest levels of pleiotropy substantially increase the variability of causal effect estimates, while the inclusion of weak IVs does not substantially affect the direction and variability of causal effect estimates in most cases. In real data analyses, we used two released versions of FinnGen GWAS summary statistics with different sample sizes as exposure GWASs to assess the influence of weak IVs. Here, the inclusion of IVs with higher exposure-association p-values resulted in weakened estimated effect sizes, particularly when the exposure GWAS sample size was small. These findings suggest that incorporating weak IVs is reasonable when the exposure GWAS sample size is large, but it poses a risk of falsely concluding null associations when the exposure GWAS sample size is small.

Read & Discuss → View Source →
09.
arXiv (CS.AI) 2026-06-17

Shattering the Autoregressive Curse: Dynamic Epistemic Entropy Orchestrated Erasable Reinforcement Learning for LLMs

Authors:
Ziliang Wang↗Kang An↗Faqiang Qian↗Jialu Cai↗Cijun Ouyang↗Yuhang Wang↗Qibing Ren↗Yichao Wu↗

arXiv:2606.17735v1 Announce Type: new Abstract: Although reinforcement learning (RL) has expanded the cognitive boundaries of large language models (LLMs), it often remains vulnerable to the autoregressive curse in long-horizon logical reasoning: small epistemic perturbations introduced early in generation can propagate irreversibly along the Markov decision process flow, triggering cascading failures that drive the reasoning trajectory toward collapse. To overcome this autoregressive cascade, in which a single early mistake can compromise all subsequent reasoning steps, we propose dynamic epistemic entropy orchestrated erasable reinforcement learning ($E^3RL$). $E^3RL$ eliminates reliance on external signals by grounding the model's endogenous local autoregressive cross-entropy as an intrinsic coordinate of epistemic uncertainty. By introducing segment-level adaptive dynamic thresholds and advantage allocation, $E^3RL$ enables the model to precisely excise localized logical defects while reusing historical key-value (KV) cache streams, thereby endowing the reasoning process with a self-healing capability. We train $E^3RL$ on the DeepMath-103k dataset. Experimental results show that $E^3RL$ reshapes the exploration efficiency of long-sequence reasoning and improves sample efficiency while maintaining linear memory overhead. On mathematical reasoning benchmarks such as AIME, $E^3RL$ achieves substantial performance gains, with the 4B and 8B parameter models surpassing previous state-of-the-art (SOTA) results by 5.349\% and 6.514\%, respectively. These findings suggest that $E^3RL$ shatters the autoregressive curse in long-sequence reasoning and establishes a theoretical and systems-level foundation for the next generation of self-healing artificial general intelligence (AGI).

Read & Discuss → View Source →
10.
arXiv (CS.AI) 2026-06-12

Humor Style Drives Laughter, Topic Shapes Acceptability: Evaluating Bilingual Personal and Political Robot-Delivered AI Jokes

Authors:
Anna-Maria Velentza↗Anne-Gwenn Bosser↗

arXiv:2606.13256v1 Announce Type: cross Abstract: Humor plays a central role in human social relationships, and recent advances in computational humor create new opportunities for integrating humor into human-robot interaction (HRI). While large language models (LLMs) can generate diverse forms of humor, it remains unclear how humor style, joke content, and language preference shape perceptions of robot-delivered humor in group settings. In this exploratory study, we employed a mixed factorial design in which participants evaluated AI-generated jokes delivered by a robot in a university classroom. We examined the effects of humor type (Affiliative, Self-Enhancing, Aggressive, Self-Defeating) and joke content (person-related vs. political) on perceived funniness and appropriateness, as well as preferred language. Results show that humor type significantly influences funniness, with Aggressive and Affiliative humor rated higher, while joke content primarily affects appropriateness, with person-related jokes preferred over political ones. Language preference was shaped by both joke content and participants' self-reported fluency and humor practices.

Read & Discuss → View Source →
11.
arXiv (CS.CV) 2026-06-18

InTrain: Intrinsic Trainability for Zero-Cost Neural Architecture Search

Authors:
Qinqin Zhou↗Fuhai Chen↗Jipeng Wu↗Zhiwei Chen↗Zhikai Hu↗Weiwei Cai↗

Training-free neural architecture search promises efficient discovery of high-performance networks without costly training. However, existing zero-cost proxies rely on fragmented heuristics that fail to capture the fundamental question: what makes an architecture trainable? This paper introduces Intrinsic Trainability (InTrain), a unified theoretical proxy that formalizes trainability as an architectural invariant emerging from two synergistic components: geometric capacity and optimization resilience. We operationalize intrinsic trainability through analysis of neural information processing. Geometric capacity is quantified via the participation ratio of activation covariance eigenspectrum, capturing the effective dimensionality of representation manifolds. Optimization resilience is measured through cumulative gradient health, assessing the robustness of backpropagation across network depth. InTrain synthesizes these dimensions through a scale-invariant multiplicative coupling, which we hypothesize is essential for capturing their synergistic, non-additive relationship. Extensive experiments on standard NAS benchmarks and search spaces demonstrate that InTrain achieves ranking correlations on par with state-of-the-art ensemble-based proxies and outperforms other single-metric methods.

Read & Discuss → View Source →
12.
arXiv (CS.LG) 2026-06-11

Learning What to Say to Your VLA: Mostly Harmless Vision Language Action Model Steering

Authors:
Hyun Joe Jeong↗Gokul Swamy↗Andrea Bajcsy↗

arXiv:2606.12299v1 Announce Type: cross Abstract: Vision-Language-Action (VLA) models provide a natural language interface to robot control, but the mapping from language to behavior is often brittle and unintuitive: semantically similar instructions can induce drastically different behaviors, while some capabilities may not be elicitable through prompting alone. As a result, both human instructions and zero-shot language models can fail to reliably steer VLAs toward successful task execution. In this work, we propose a framework that interactively searches for language sequences that improve closed-loop VLA task performance, distills these sequences into a test-time language feedback policy (LFP), and learns an improvement head that predicts when language steering will improve performance. We conformalize this improvement head to prevent harmful steering interventions, where the LFP decreases task performance relative to the original instruction on out-of-distribution scenarios. Crucially, our approach operates on arbitrary frozen pre-trained VLAs, requiring neither access to the original training distribution nor fine-tuning of the underlying model. On seen environments, our conformalized LFP improves base VLA performance by 24.7% in simulation and 65.0% in hardware. On visual and semantic perturbations, our conformalized LFP has strong harmlessness guarantees, and produces recovery behaviors not observed with open-loop prompting.

Read & Discuss → View Source →
13.
arXiv (CS.CL) 2026-06-17

Are you speaking my languages? On spoken language adherence in multimodal LLMs

Authors:
Hyungwon Kim↗Kandarp Joshi↗Lillian Zhou↗Pavel Golik↗Petar Aleksic↗

While Large Language Model (LLM) based Automatic Speech Recognition (ASR) enables seamless multilingual use, models often misidentify the output language, compromising transcription fidelity and downstream application quality. To preserve flexibility and code-switching capabilities, we propose a soft prompting approach that hints at potential spoken languages without strictly constraining the output. We formally define this challenge as a lack of language adherence, introduce a novel metric to quantify violations, and evaluate three mitigation strategies: (1) zero-shot prompting for robust guidance under uncertainty, (2) supervised fine-tuning (SFT) to improve prompt adherence, and (3) Chain-of-Thought (CoT) reasoning to enforce adherence during decoding. We present a comparative analysis of these methods across multiple languages, evaluating effectiveness in reducing the language violation while maintaining overall ASR performance. Finally, we discuss trade-offs to guide strategy selection under various compute constraints.

Read & Discuss → View Source →
14.
arXiv (CS.CV) 2026-06-12

Trajectory-Level Redirection Attacks on Vision-Language-Action Models

Authors:
Gokul Puthumanaillam↗Vardhan Dongre↗Pranay Thangeda↗Hooshang Nayyeri↗Dilek Hakkani-T\"ur↗Melkior Ornik↗

Vision-language-action (VLA) policies bring natural language into closed-loop robot control, enabling robots to execute manipulation tasks directly from text instructions. The same interface gives text a recurring role in control because the prompt is reused at every replanning step, and each prompt-conditioned action changes the future observations on which the policy acts. Existing VLA attacks study adversarial prompts that elicit targeted low-level actions or make such actions persist across changing images. We identify a stronger trajectory-level failure mode: a prompt that still $appears$ to specify the intended task but redirects the final physical outcome. We mathematically formalize this setting as $command-preserving trajectory redirection$, a prompt-only threat model in which the attacker chooses one prompt before the episode, all policy and environment components remain fixed, and the prompt must stay close to the benign instruction while omitting target words and correction language. To find such prompts, we introduce an on-policy prompt search method that uses rollouts to discover perturbations whose closed-loop behavior tracks a target task while satisfying the command-preserving constraints. Experiments in simulation and on hardware show that near-benign prompt perturbations can redirect VLA rollouts to attacker-specified targets. These results expose a trajectory-level vulnerability in VLA instruction grounding: text that appears to preserve the intended command can still give an adversary control over the robot's final physical outcome. Project website: https://vla-redirection-attack.github.io/

Read & Discuss → View Source →
15.
arXiv (math.PR) 2026-06-15

On the Poisson Follower Model

Authors:
Natasa Dragovic↗Francois Baccelli↗

arXiv:2309.04864v5 Announce Type: replace Abstract: We introduce a stochastic geometry dynamics inspired by opinion dynamics that captures the essence of modern asymmetric social networks with leaders and followers. Points in the Euclidean space represent opinions, and the leader of an agent is the one with the closest opinion. In this dynamics, each follower updates its opinion by halving the distance to its leader. We demonstrate that this simple dynamics and its iterations exhibit several interesting purely geometric phenomena related to the evolution of leadership and opinion clusters, which resemble those observed in social networks. We also show that when the initial opinions are randomly distributed as a stationary Poisson point process, the spatial frequency of each of these phenomena can be expressed through an integral geometry formula involving semi-algebraic domains. Finally, we analyze numerically the limiting behavior of this follower dynamics. In the Poisson case, the agents fall into two categories: ultimate followers, who continue updating their opinions indefinitely, and ultimate leaders, who adopt a fixed opinion after a finite time. Spatial discrete event simulations support all our findings.

Read & Discuss → View Source →
16.
arXiv (CS.CL) 2026-06-12

X-MADAM-RAG: Diagnosing and Handling Chinese-English Evidence Conflict in Retrieval-Augmented Generation

Authors:
Yongqi Kang↗Yu Fu↗Yong Zhao↗

Retrieval-augmented generation (RAG) systems may receive evidence that is not merely noisy but mutually contradictory. This issue becomes particularly salient in multilingual settings, where retrieved Chinese and English evidence may support incompatible answer candidates. We study this problem through X-RAMDocs-ZHEN, a controlled Chinese-English benchmark derived from RAMDocs for diagnosing evidence conflict in RAG. The benchmark contains 300 examples across six balanced conditions, including monolingual support, bilingual agreement, reversed conflict directions, and conflict with optional noise. We further examine X-MADAM-RAG, an interpretable pipeline that decomposes evidence handling into per-document candidate extraction, visible-evidence repair, deterministic candidate grouping, and conflict-aware aggregation. On the original controlled benchmark with Qwen2.5-7B-Instruct, X-MADAM-RAG achieves 0.9667 strict accuracy and 0.9767 conflict-aware success, outperforming an evidence-normalized single-call baseline. However, a zero-call rule-only extractor reaches 1.0000 on the same benchmark, revealing strong template regularity. To probe this limitation, we construct a deterministic naturalized stress test that removes explicit answer templates while preserving candidate strings. On its 100-sample subset, rule-only extraction falls to 0.0000, but X-MADAM-RAG also drops to 0.3000 strict accuracy, below both naive and evidence-normalized baselines. A privileged oracle remains perfect, indicating that document-level extraction is the main bottleneck. These findings position X-RAMDocs-ZHEN and X-MADAM-RAG as diagnostic tools for controlled evidence conflict rather than as evidence of general hallucination detection or robustness to natural retrieval.

Read & Discuss → View Source →
17.
arXiv (CS.CV) 2026-06-16

Sex-based Network-Specific Differences in Connectomes: A Krakencoder-Based Analysis

Authors:
Vibhashree S H↗Debanjali Bhattacharya↗Vamshi Krishna Kancharla↗Neelam Sinha↗

This study examines how deficiencies in one brain connectome modality propagate to the other, using the Krakencoder as a simulation framework. Structural and functional connectomes from 702 healthy participants in the Human Connectome Project were analyzed, with the impact of each of the Yeo-7 functional networks assessed separately. Seven scenarios were considered, each involving the removal of a single network while the remaining networks were preserved. The resulting perturbations in cross-modal predictions were quantified using three complementary metrics: KL divergence on eigenvalue spectra, Frobenius norm, and Wasserstein distance. In addition, the persistence of sex-specific information within the predicted connectomes was evaluated. Across all metrics and both prediction directions, the Default Mode Network produced the largest perturbations, whereas the Somatomotor network yielded the smallest. Sex differences in network-level perturbation signatures were subtle, with the best result being an accuracy of 66.09% from connectomes predicted under network-removal conditions. In contrast, connectomes predicted from intact inputs achieved substantially higher sex classification accuracy, reaching up to 84.76%. These findings confirm that full predicted connectomes retain considerably more sex-discriminative information than perturbation-derived signatures alone.

Read & Discuss → View Source →
18.
medRxiv (Medicine) 2026-06-23

Differential Recovery Trajectories of Emergency Otolaryngologic Conditions across the COVID-19 Pandemic: A Six-year Longitudinal Study from an Urban Emergency Center

Authors:
Ogawa↗

Objective: The COVID-19 pandemic markedly altered social activity patterns, healthcare utilization, and the epidemiology of infectious diseases. However, its long-term impact on emergency otolaryngologic conditions remains incompletely understood. This study investigated long-term trends in emergency otolaryngologic conditions before, during, and after the COVID-19 pandemic using comprehensive data from a large urban emergency clinic in Osaka, Japan. Methods: All new otolaryngologic outpatients who visited the Chuo Emergency Medical Clinic (CEMC) in Osaka City between 2019 and 2024were retrospectively analyzed. Annual trends in absolute numbers and relative proportions of emergency otolaryngologic conditions were examined by anatomical region and disease category, using 2019 as the pre-pandemic baseline. Results: A total of 99,324 new otolaryngologic outpatients were analyzed. Overall emergency visits declined sharply to approximately half of baseline in 2020, followed by a gradual but incomplete recovery toward pre-pandemic levels by 2024. Most anatomical categories declined to 45-61% of baseline in 2020 and exhibited gradual yet incomplete recovery through 2023; in stark contrast, laryngeal conditions diverged sharply, surging beyond pre-pandemic levels after 2022. Acute infectious otorhinolaryngologic diseases fell to 23-50% of baseline in 2020 and showed variable recovery (69-103%) by 2024. Notably, laryngitis exceeded the baseline, reaching 132% in 2023, whereas epiglottic edema exhibited only a transient increase approaching the baseline in 2021. Non-infectious emergency conditions generally showed only a marginal decrease in 2020 and remained relatively stable throughout the study period, except for sudden sensorineural hearing loss (SSNHL), which dropped sharply to 39% of the baseline in 2020 and remained persistently reduced through 2024. Traumatic emergencies declined variably to 53-81% of the baseline in 2020, followed by an incomplete recovery, reaching only 55-69% by 2024. Conclusion: Emergency otolaryngologic conditions demonstrated heterogeneous recovery trajectories following the COVID-19 pandemic. While most infectious and traumatic conditions gradually but incompletely normalized, laryngeal conditions showed a distinct post-pandemic surge, and SSNHL remained persistently suppressed. These findings reveal heterogeneous, condition-specific recovery trajectories that reflect both genuine shifts in community pathogen burden, true traumatic incidence, and persistent alterations in healthcare-seeking behaviors, insights essential for resource allocation during future public health emergencies.

Read & Discuss → View Source →
19.
arXiv (CS.AI) 2026-06-15

Distributional Biases in Post-Training: A Markovian Analysis of Reasoning Trajectories

Authors:
Dake Bu↗Wei Huang↗Andi Han↗Atsushi Nitanda↗Bo Xue↗Qingfu Zhang↗Hau-San Wong↗Taiji Suzuki↗

arXiv:2511.07368v3 Announce Type: replace-cross Abstract: Foundation models exhibit broad knowledge but limited task-specific reasoning, motivating post-training strategies such as RL with verifiable rewards (RLVR) and test-time scaling (TTS). While recent work highlights the role of exploration in improving pass@K, empirical evidence points to a paradox: RLVR and ORM/PRM typically reinforce existing paths rather than expanding the reasoning scope, raising the question of why exploration helps if no new patterns emerge. To reconcile this paradox, we adopt the perspective of Kim et al. (2025), viewing easy (e.g., simplifying a fraction) versus hard (e.g., discovering the some symmetry) reasoning steps as low versus high probability Markov transitions. In this tractable model, pretraining corresponds to tree-graph discovering, while post-training corresponds to CoT reweighting. We provably show that, both RLVR and ORM/PRM would favor heavily to several high-probability paths, and thereby forget rare-but-crucial CoTs. Building on this, we further prove that exploration strategies such as rejecting easy instances and KL regularization help preserve rare CoTs. Empirical simulations corroborate our theoretical results.

Read & Discuss → View Source →
20.
arXiv (CS.AI) 2026-06-16

Poster: EdgeCitadel – Hybrid NATS-MQTT Orchestration for Edge Multi-Agent Systems

Authors:
Zhonghao Zhan↗Yefan Zhang↗Hamed Haddadi↗

arXiv:2606.14710v1 Announce Type: cross Abstract: Edge-resident AI agents increasingly span home servers, IoT hubs, laptops, and phones, yet their coordination stacks still assume cloud-style transports or a central relay. We present EdgeCitadel, an edge multi-agent orchestration platform built around a single NATS 2.10 server with the built-in MQTT adapter. The design combines MQTT connectivity for heterogeneous agents, JetStream-backed persistence and replay for backend services, direct peer delegation over a shared subject namespace, and a passive aggregator that visualizes and stores traffic without sitting on the delivery path. Our poster highlights the migration from MQTT relay prototypes (common in IoT communication) to the current hybrid architecture and demonstrates a working cross-device testbed spanning ARM64, x64, and Android clients.

Read & Discuss → View Source →
21.
arXiv (CS.LG) 2026-06-18

MOLAR: Learning Multimodal Molecular Representations from Noisy Labels

Authors:
Yingxu Wang↗Kunyu Zhang↗Nan Yin↗Yu Li↗Eran Segal↗

arXiv:2606.18390v1 Announce Type: new Abstract: Motivation: Noisy labels are a common challenge in molecular property prediction because molecular annotations are often obtained from assays, curated databases, or weak annotation pipelines rather than directly observed clean biological states. Treating recorded labels as reliable supervision can cause models to memorize corrupted observations and learn misleading molecular evidence. In multimodal molecular representation learning, this issue can be amplified by graph-text fusion or alignment, which may propagate label-induced errors across modalities. Results: We propose MOLAR, a noise-aware framework for learning multimodal molecular representations from noisy labels. MOLAR separates latent clean-property inference from recorded-label observation: graph and text views contribute residual evidence to a clean-property distribution, and a categorical label-observation channel maps this distribution to recorded labels for training. This formulation derives posterior label reliability and modality-specific molecular evidence from the model. Experiments on naturally noisy molecular benchmarks and controlled label-flipping benchmarks show that MOLAR consistently outperforms representative baselines. Visualization analyses further show that MOLAR provides interpretable reliability and modality-evidence diagnostics.

Read & Discuss → View Source →
22.
arXiv (CS.AI) 2026-06-16

Surprise-Guided MergeSort: Budget-Efficient Human-in-the-Loop Ranking via Adaptive Comparison Scheduling

Authors:
Yujin Park↗Haejun Chung↗Ikbeom Jang↗

arXiv:2606.15623v1 Announce Type: cross Abstract: Pairwise comparison is the gold standard for subjective ranking tasks; however, exhaustive annotation requires a massive number of human comparisons ($O(n^2)$). While sorting-based methods have reduced this burden to $O(n\log n)$, they still require expensive human judgment for every single comparison. To further improve annotation efficiency, we propose leveraging a Vision-Language Model (VLM) not as an annotator replacement, but as a question prioritizer to identify which comparisons genuinely require human judgment. The proposed Surprise-Guided MergeSort (SGS) framework achieves this through three integrated components: (1) a bottom-up MergeSort scheduler that structures comparisons and exploits transitivity, (2) a composite Surprise Scorer – combining position-bias-cancelled VLM confidence, Elo gap, and vote entropy – to quantify comparison ambiguity, and (3) an adaptive budget allocator that routes high-surprise pairs to humans while automating low-surprise pairs via transitivity inference. Validation was conducted on six diverse benchmarks spanning text similarity (STS-B, BIOSSES, SICKR-STS) and image quality assessment (KonIQ-10k, TID2013, LIVE Challenge). SGS effectively identified and skipped up to 535 non-informative comparisons per session. Consequently, it achieved Kendall's $\tau{\times}100$ improvements of $+6$ to $+12$ over Active Elo under the same total budget. These results demonstrate that combining VLM-guided surprise metrics with algorithmic sorting provides a generally consistent accuracy-efficiency trade-off across diverse domains.

Read & Discuss → View Source →
23.
arXiv (CS.CL) 2026-06-11

AI Coding Agents in Social Science: Methodologically Diverse, Empirically Consistent, Interpretively Vulnerable

Authors:
Meysam Alizadeh↗Fabrizio Gilardi↗Mohsen Mosleh↗Enkelejda Kasneci↗

The deployment of LLM-based agents in scientific analysis raises opposing concerns: that agents may reduce methodological diversity, or that they may amplify the analytic flexibility through which researchers reach motivated conclusions. We argue these worries target two empirically separable layers: a design layer of methodological choices, and a verdict layer in which a decision rule maps estimates to a substantive claim. We test both by running 20 independent executions of Claude Code and Codex on a prominent immigration and social-policy against a many-analysts human baseline. At the design layer, Codex matches human methodological diversity and Claude Code produces nearly three times as many specifications; both agents' effect estimates remain broadly aligned with the human consensus, and no agent model exactly matches any human model. A prompt-induced anti-immigration researcher prior reorganizes each agent's methodological decisions but, unlike for biased human analysts in the same data, does not shift aggregate estimates or final verdicts; nor do agents reroute along the methodological axes humans use to bias their estimates. At the verdict layer, an explicit confirmatory prompt flips Claude Code's verdicts from 10% to 90% support while leaving its coefficient distribution essentially unchanged, operating through rule omission rather than rule softening. AI agents can rival or exceed human methodological diversity at the design layer while remaining vulnerable at the verdict layer. In our setting, the locus of AI bias is not estimation but interpretation.

Read & Discuss → View Source →
24.
arXiv (CS.AI) 2026-06-12

Representing Time Series as Structured Programs for LLM Reasoning

Authors:
Jaeho Kim↗Changhun Oh↗Seokhyun Lee↗Irina Rish↗Changhee Lee↗

arXiv:2606.12481v1 Announce Type: cross Abstract: Large language models (LLMs) have demonstrated strong reasoning and instruction-following capabilities, making them potentially powerful tools for time-series analysis. However, time series lie outside their native textual modality, raising a fundamental question: how should time series be represented so that LLMs can reason about them effectively? Existing work typically serializes raw numerical sequences or fine-tunes pre-trained LLMs on time-series data. These approaches place the burden of extracting temporal structure directly on the LLM, creating a modality mismatch that often degrades performance on long sequences and introduces substantial computational overhead. In this work, we introduce Time-Series-to-Structured-Program representation (T2SP), a deterministic, training-free method that represents a time series as a structured symbolic program. T2SP decomposes time series into trends, periods, and salient events, expressing them in a program-friendly format aligned with the textual and code-like modalities on which LLMs are natively trained. By shifting temporal-structure extraction from the model to the representation itself, T2SP enables off-the-shelf LLMs to leverage their existing reasoning capabilities for time-series understanding. We evaluate T2SP on three reasoning tasks – editing, captioning, and question answering – where it consistently improves performance, reduces reasoning time, and lowers failure rates compared with raw-string representations. Our results demonstrate that T2SP provides an effective interface between time series and LLMs.

Read & Discuss → View Source →
25.
arXiv (CS.LG) 2026-06-15

Trust but Verify: Mitigating Medical Hallucinations via Post-Hoc Adversarial Auditing and Multi-Agent Feedback Loops

Authors:
Muhammad Osama↗Maheera Amjad↗Zartasha Mustansar↗Arslan Shaukat↗Muhammad U. S. Khan↗

arXiv:2606.14149v1 Announce Type: new Abstract: Large Language Models (LLMs) are increasingly deployed in healthcare settings, yet their tendency to hallucinate poses risks when clinical decisions are involved. This study examine whether LLMs recommend recently banned or withdrawn pharmaceuticals when answering clinical questions and tests an agent-based method for reducing such errors. We developed a five-agent "Trust but Verify" system using a single LLM backbone. To measure regulatory knowledge obsolescence, we created an adversarial dataset of 103 clinical MCQs where historically correct answers now refer to banned substances. This scale ensures statistical significance across various therapeutic classes. We evaluated three open-access model families (GPT-OSS, Llama-3, Falcon-3) under vanilla and agentic conditions. Performance was measured via pointwise score, label accuracy, Hallucination Error Rate (HER), and Component Fidelity (CF) score. We also observed clinical safety regression in proprietary models. In default configurations, all models showed high hallucination rates, consistently selecting banned drugs that matched training data patterns. Our proposed agentic architecture reduced HER by approximately 53% across models. Pointwise scores shifted from -0.25 (unsafe recommendation) toward 0.0 (appropriate refusal). The safety audit intercepted dangerous outputs even when models' parametric knowledge favored the banned substance. The proposed multi-agent framework offers a model-agnostic method for enforcing regulatory compliance that prioritizes patient safety over fluent text generation. Our work demonstrates a practical approach for deploying autonomous AI systems in safety-critical healthcare settings. It shows how real-time regulatory data can be integrated into LLM pipelines to support clinical decision-making.

Read & Discuss → View Source →