Academic Intelligence · Curated Daily

探索全球前沿学术脉络

AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达,利用大语言模型自动生成交叉领域文献分析简报。

01.
arXiv (CS.LG) 2026-06-15

A theoretical model for task routing in mixture-of-expert transformers

arXiv:2606.14398v1 Announce Type: new Abstract: Mixture-of-experts (MoE) layers enable the scaling of transformer models while keeping the inference compute fixed. While task-expert specialization has been observed in empirical studies of frontier MoE transformer models, existing theoretical work analyzes this using continuous mixture models that cannot be used to model natural language effectively. An important open question is to theoretically explain task-expert specialization in transformer MoE models using discrete models of language. To address this, we represent structured knowledge via syntactic templates and finite key-value dictionaries, and prove formally that a single-layer MoE transformer can encode knowledge by using experts that specialize in the corresponding tasks. Our construction shows how queries are routed to unique, task-specific experts whose size depends solely on the intrinsic complexity of the given task (i.e. the combined size of its syntactic templates and factual dictionary). Our construction provides a theoretical support for empirical results on localized knowledge circuits in MoE models. We support our theoretical findings with experiments evaluating model performance under varying MoE loss functions.

02.
arXiv (CS.CL) 2026-06-25

A Red Teaming Framework for Large Language Models: A Case Study on Faithfulness Evaluation

Large language models (LLMs) have demonstrated remarkable performance across natural language processing tasks, yet their deployment in high-stakes applications raises critical concerns regarding reliability, safety, and trustworthiness. In this paper, we present a red teaming framework that systematically uncovers vulnerabilities in LLM outputs. Our approach employs a novel multi-role architecture comprising target, attacker, and jury models. The attackers generate increasingly effective adversarial prompts while the jury rigorously evaluates response accuracy and consistency across tasks. In a case study, our strategy proved particularly effective at exposing unfaithfulness in LLM responses. Exploitative adversarial prompts increased the attack success rate by up to 7.9% in question-answering tasks, revealing weaknesses in reliability. The approach identifies how structural constraints in summarization can shape vulnerability patterns, with format limitations yielding measurable gains in faithfulness, and shows that architectural design choices typically outweigh parameter scaling in determining model safety. The framework's key strength is its adaptability across evaluation tasks, from English question-answering to Arabic summarization, enabling comprehensive comparison of model vulnerabilities. While it excels at comparing cross-model and cross-linguistic vulnerabilities, it faces challenges in fully automating adversarial prompt generation across languages. Our experiments also reveal limitations in detecting subtle forms of unfaithfulness that do not manifest as explicit factual contradictions, particularly across linguistic contexts. Overall, this architecture provides both actionable insights into current LLM vulnerabilities and a scalable methodology for ongoing safety evaluation as models evolve.

03.
arXiv (CS.LG) 2026-06-16

Learning Policy from a Single Trajectory in Average-Reward Markov Decision Process

arXiv:2606.16729v1 Announce Type: new Abstract: While there is an extensive body of work characterizing the sample complexity of discounted cumulative-reward MDPs, finite sample analyses for average-reward MDPs have been limited, and most existing works rely on restrictive assumptions such as ergodicity or access to a generative model. In this work, we establish the first finite sample complexity guarantees from a single trajectory for weakly communicating average-reward MDPs. To this end, we study the dynamics of a single trajectory in weakly communicating MDPs and based on this analysis, we develop novel model-free methods. Notably, our value-based and policy-based methods provide finite sample complexity guarantees of $\widetilde{O}(1/\varepsilon^2)$ and $\widetilde{O}(1/\varepsilon^4)$ from a single trajectory in weakly communicating MDPs, respectively. Furthermore, we introduce the first model-free method that requires no prior knowledge of problem-dependent quantities for communicating MDPs.

04.
arXiv (quant-ph) 2026-06-24

Faster algorithm for achieving minimal-size quantum decision diagrams

arXiv:2606.24789v1 Announce Type: new Abstract: The decision diagram (DD) data structure enables fast linear-algebra calculations by bringing vectors into a normal form and subsequently merging equivalent ones, yielding a minimally-sized DD modulo the equivalence relation. A fruitful application area is quantum-circuit simulation, where the vectors represent quantum states. The Local Invertible Map Decision Diagram (LIMDD) type, merges LIM-equivalent (typically Pauli-gate equivalent) vectors, can efficiently simulate Clifford circuits as well as some high-T-count circuits, and has theoretically been proven exponentially faster for simulation than other well-developed data structures, including other common DD variants. However, these exponential advantages have not fully materialized yet in existing implementations, for which the normal-form procedure, which is a highly complex algorithm, is either absent or only partially implemented. We here present a novel normal-form algorithm for Pauli-LIMDDs, achieving a worst-case speedup from $O(n^3)$ to $O(n^2)$ for an $n$-qubit DD node with a single child node while keeping the $O(n^3)$ run time in case of two distinct children nodes. We implement the algorithm as part of QolDDer, our Pauli-LIMDD simulator for quantum circuits, written from scratch in C/C++. The implementation realizes the theoretically-proven advantages of Pauli-LIMDDs on Clifford circuits, is significantly faster than the existing LIMDD simulators on such circuits, and on a public quantum-circuit data set often outperforms them by an order of magnitude. In the future, we envision that our work will enable further application and development of LIMDD variants, not only for quantum design tasks, but also for analysis of linear-algebra-based systems in general.

05.
arXiv (CS.AI) 2026-06-17

Unlocking LLM Code Correction with Iterative Feedback Loops

arXiv:2606.17514v1 Announce Type: cross Abstract: Large Language Models have shown remarkable capabilities in code generation. However, most existing evaluations focus only on single-attempt accuracy and overlook the iterative refinement process that is central to real-world programming. This study presents a systematic investigation of LLMs' ability to rectify their own code through execution feedback. Using real-world programming problems across four models and two major programming languages, this study evaluates performance using iterative refinement framework where LLMs receive compiler error messages and testcase feedback after each attempt. This study introduces metrics to evaluate code failures, analyze rectification patterns, and compare the effectiveness of reasoning and non-reasoning models, offering actionable insights into both the understanding and practical application of feedback loops in LLM-driven code generation systems. Results show that reasoning models consistently improve over iterations, substantially outperforming non-reasoning models in leveraging feedback, while syntactic and runtime errors are far more tractable than logical or algorithmic failures.

06.
medRxiv (Medicine) 2026-06-22

Efficacy and safety of semaglutide for obesity and hyperphagia in adults with Prader-Willi syndrome

Context: Prader-Willi syndrome is a genetic neurodevelopmental disorder characterized by hyperphagia and early-onset obesity from hypothalamic dysfunction with endocrinopathies and learning disability. Management is challenging with strict control of the food environment needed. While newer glucagon-like peptide-1 receptor agonists, such as semaglutide, have efficacy in non-PWS obesity, there have been limited case reports in PWS. Objective/Design/Setting: Retrospective records review of 12 adults with PWS and overweight/obesity treated with semaglutide at a UK academic hospital centre specialist clinic. Patients: mean +/- SD age 28.3 +/- 10.1 years, 83% female, BMI 46.6 +/- 8.2kg/m2, 75% type 2 diabetes mellitus. Intervention: Median follow-up 17.2 months (range 8.7-36.1) with median semaglutide dose 2.4mg once weekly (1.0-2.4). Results: Although there was no significant weight loss on semaglutide, there was stabilisation of the weight gain prior to treatment over previous 12.4 months (7.6-23.0) (post -3.1 +/- 9.9% vs. pre +5.7 +/- 5.6%: d -0.72, P=0.037). There was a significant decrease in hyperphagia on semaglutide from hyperphagia questionnaire for clinical trials (n=11, -7.3 +/- 6.1 (max 36), d -1.19, P=0.003), having been stable before treatment. HbA1c improved in those with elevated baseline levels (n=6, -4.2 +/- 4.9%, d -0.74, P=0.13). Mild gastrointestinal side effects were seen in 25% but did not lead to discontinuation. Conclusions: In adults with PWS, semaglutide produced weight maintenance, reduced hyperphagia, and improved glycaemic control, with good tolerability. Larger placebo-controlled trials are needed to confirm these findings in adults and adolescents with PWS, especially in those without T2DM, where efficacy may be greater.

07.
arXiv (CS.CV) 2026-06-19

CoMo: Learning Continuous Latent Motion from Internet Videos for Scalable Robot Learning

Unsupervised learning of latent motion from Internet videos is crucial for robot learning. Existing discrete methods generally mitigate the shortcut learning caused by extracting excessive static backgrounds through vector quantization with a small codebook size. However, they suffer from information loss and struggle to capture more complex and fine-grained dynamics. Moreover, there is an inherent gap between the distribution of discrete latent motion and continuous robot action, which hinders the joint learning of a unified policy. We propose CoMo, which aims to learn more precise continuous latent motion from internet-scale videos. CoMo employs an early temporal difference (Td) mechanism to increase the shortcut learning difficulty and explicitly enhance motion cues. Additionally, to ensure latent motion better captures meaningful foregrounds, we further propose a temporal contrastive learning (Tcl) scheme. Specifically, positive pairs are constructed with a small future frame temporal offset, while negative pairs are formed by directly reversing the temporal direction. The proposed Td and Tcl work synergistically and effectively ensure that the latent motion focuses better on the foreground and reinforces motion cues. Critically, CoMo exhibits strong zeroshot generalization, enabling it to generate effective pseudo action labels for unseen videos. Extensive simulated and real-world experiments show that policies co-trained with CoMo pseudo action labels achieve superior performance with both diffusion and auto-regressive architectures.

08.
arXiv (quant-ph) 2026-06-17

A Lindbladian for holographic Brownian motion

arXiv:2606.17909v1 Announce Type: cross Abstract: We derive a Lindbladian description of holographic Brownian motion in the high-temperature regime. Starting from the influence functional for a trailing string endpoint, we identify the corresponding quantum master equation and prove that it is completely positive and trace-preserving. We determine the coefficients of the Lindbladian explicitly for two holographic backgrounds: the BTZ black hole and the AdS$_5$ black brane, restricting in the latter case to the endpoint fluctuation along the $x^1$-direction. We then analyze the time evolution of phase-space moments, energy relaxation, and steady states.

09.
arXiv (CS.AI) 2026-06-24

CompressKV: Semantic-Retrieval-Guided KV-Cache Compression for Resource-Efficient Long-Context LLM Inference

arXiv:2606.24467v1 Announce Type: new Abstract: Long-context large language model (LLM) inference is increasingly constrained by the memory footprint and decoding cost of key-value (KV) caches, limiting sustainable deployment on resource-constrained hardware. Existing KV cache eviction methods typically apply heuristic token scoring over all heads in GQA-based LLMs. These methods ignore the different functionalities of attention heads, leading to the eviction of critical tokens and thus degrading the performance of LLMs. To address this issue, we propose CompressKV, a resource-efficient KV-cache compression framework for GQA-based LLMs. Instead of aggregating attention scores from all heads, CompressKV identifies Semantic Retrieval Heads (SRHs) that capture both the initial and final tokens of a prompt and semantically important mid-context evidence, and uses them to select tokens whose KV pairs should be retained. Furthermore, CompressKV allocates cache budgets across layers according to offline estimates of layer-wise eviction error. Experiments on LongBench and Needle-in-a-Haystack show that CompressKV consistently outperforms existing KV-cache eviction methods across memory budgets. Notably, it preserves over 97\% of full-cache performance using only 3\% of the KV cache on LongBench question-answering tasks and achieves 90\% accuracy with just 0.7\% KV storage on Needle-in-a-Haystack. These results demonstrate an improved resource–performance trade-off for long-context LLM inference. Our code is publicly available at: https://github.com/TUDa-HWAI/CompressKV

10.
arXiv (CS.CL) 2026-06-11

WorldReasoner: Evaluating Whether Language Model Agents Forecast Events with Valid Reasoning

Forecasting real-world events requires language-model agents to reason under uncertainty from incomplete, time-bounded information. Yet evaluating whether agents genuinely forecast requires more than final-answer accuracy: a model may be correct by recalling memorized training facts, citing fabricated evidence, or producing an unsupported causal story. We present WorldReasoner, an evaluation framework for temporally valid event forecasting. Each task gives an agent a resolved forecasting question, a simulated forecast date, and access only to evidence available before that date; after resolution, the framework scores the submitted probability, cited evidence, and optional causal event graph. WorldReasoner reports three complementary axes: outcome quality against resolved answers, evidence quality over cited sources, and reasoning quality against post-resolution hindsight graphs. The benchmark is built by an agentic construction pipeline that generates forecasting questions, collects time-stamped evidence, and builds hindsight reference graphs at scale, yielding 345 resolved tasks derived from 14,141 articles with graphs covering 8,087 extracted events. Across six controlled agent settings, temporally valid retrieval is the strongest driver of outcome accuracy; causal graph construction improves key-event recovery; and correct graph-enabled forecasts are more strongly grounded in key events and relevant sources, yet agents still struggle to convert grounded evidence into calibrated probabilities.

11.
medRxiv (Medicine) 2026-06-11

Ferritin across long-term conditions in England: cross-sectional primary care study

Background Iron deficiency (ID) is a readily treatable condition once identified. Ferritin is the primary diagnostic marker, but cut-offs vary and inflammation complicates interpretation in patients with long-term conditions (LTCs). Aim To describe ferritin distribution and the prevalence of threshold-defined low ferritin in adults with and without LTCs in primary care. Design and setting Cross-sectional observational study using routinely collected electronic health records from a national primary care database in England (1st January 2015 to 31st December 2021). Method Adults with >1 ferritin test in Clinical Practice Research Datalink (CPRD) Aurum were included. LTCs were identified using validated primary-care code lists. Outcomes included ferritin distribution and threshold-defined ID prevalence using World Health Organization (WHO) (

12.
medRxiv (Medicine) 2026-06-24

Computational Decomposition of New Memory Failure in Alzheimer's Disease Through a Hippocampal Cortical Consolidation Bottleneck Model

Alzheimer's disease (AD) is clinically marked by difficulty retaining newly learned information, yet routine memory scores often conflate poor initial encoding with failure to stabilise information after encoding. This ambiguity limits the mechanistic interpretability of cognitive assessment during the transition from mild cognitive impairment to AD. Here we propose a Hippocampal Cortical Consolidation Bottleneck (HCCB) model to computationally separate these two components of new memory failure. The model represents newly presented information as a rapidly formed hippocampal trace and a slowly stabilised cortical trace, predicting a residual bottleneck when delayed recall falls below the level expected from immediate recall. We operationalised this prediction as Consolidation Bottleneck Index*(CBI*), a cognitively normal reference normalised residual index, and evaluated it using Alzheimer's Disease Neuroimaging Initiative (ADNI) cognitive and MRI data, with independent dynamical support from OpenNeuro EEG. Simulations showed recent memory vulnerability when hippocampal vulnerability exceeded cortical vulnerability. In ADNI, CBI* increased from cognitively normal participants to mild cognitive impairment nonconverters, reached Alzheimer like levels in mild cognitive impairment converters, and was associated with hippocampal atrophy. CBI* added minimal discrimination beyond established clinical and structural predictors, supporting its role as a mechanistic phenotype rather than a replacement prognostic model. OpenNeuro EEG further showed increased neurodynamic rigidity in AD. Our findings provide a computational framework for quantifying failed stabilisation of newly encoded information in AD progression.

13.
medRxiv (Medicine) 2026-06-17

Differential Determinants of Past Behavior and Future Intention Regarding Voluntary Blood Donation: A Cross-Sectional Study of Knowledge, Attitudes, and Practices in Qingdao, China

Background A persistent gap between motivation and action threatens voluntary blood supply. This study examined the publics knowledge, attitudes, and practices (KAP) regarding blood donation, with a particular focus on identifying the different determinants of past blood donation behavior and future willingness to donate. Methods Convenience sampling was used to conduct a cross-sectional survey among 1,058 eligible people in Qingdao, China, between July and November 2025. Data were collected via a self-designed KAP questionnaire. To find independent characteristics linked to previous behavior and future intention, respectively, multivariable binary logistic regression was used. Results Overall, 37.0% of participants (n=391) had a lifetime donation history, while 39.2% (n=415) intended to donate in the next 12 months. Past behavior was positively associated with older age (36-45 years: OR=6.84; 95% CI: 3.21-14.58), higher education (OR=2.06; 95% CI: 1.33-3.17), and interpersonal interaction channels (OR=1.45; 95% CI: 1.01-2.09) but hindered by safety concerns (OR=0.23; 95% CI: 0.16-0.34). Conversely, future intention was positively correlated with male sex (OR=1.69; 95% CI: 1.24-2.29), prior donation history (OR=2.69; 95% CI: 1.87-3.86), having family members or friends in need of blood (OR=2.75; 95% CI: 1.96-3.85), and traditional media exposure (OR=3.33; 95% CI: 2.18-5.10). Higher education was adversely correlated with future intention (OR=0.55; 95% CI: 0.38-0.79). Conclusion There is a substantial disparity between donation motivation and action. The determinants of past behavior and future intention are asymmetric, suggesting that stage-specific interventions are required, using social mobilization for initiating first-time donations, while employing family reciprocity and authoritative communication to sustain long-term engagement.

14.
medRxiv (Medicine) 2026-06-24

Self-administered computerized cognitive training for cognitive deficits in individuals with metabolic syndrome: a randomized controlled trial

Background: Metabolic syndrome (MetS) has been associated with cognitive decline. Considering its increasing prevalence worldwide, the goal of this study was to evaluate the feasibility and efficacy of a short-term, self-administered computerized cognitive training programme in individuals with metabolic syndrome and low cognitive performances. Methods: Thirty six participants, aged 40-72 years (mean age: 57.8 years), were randomly assigned to the cognitive training or the passive control group. The cognitive training component of Long Lasting Memories (LLM) Care was used as an interactive software to enhance participants' cognitive functions. Up to 24 sessions, each lasting 45 minutes, were self-administered at home twice per week for 3 months. Thorough cognitive assessments with were performed at baseline (randomization), at the end of intervention, and 12 months after baseline. The primary outcome was performance at nine neuropsychological tests, and the secondary outcome was a self-reported questionnaire assessing everyday functional abilities. Primary analyses were performed employing mixed-effect models using the intention-to-treat principle. Results: Low adherence was observed in the study, as only 9 participants (50%) completed at least 8 sessions of the cognitive training programme (range 9-24 sessions, median 15 sessions). No statistically significant effect of the cognitive training programme on performance in neuropsychological tests or everyday functioning was found. At the end of the 3-month intervention programme, effect for visual memory enhancement in immediate ({beta} = 1.58, 95% CI = -1.84 to 4.99, Cohen's d = 0.39) and delayed recall ({beta} = 2.17, 95% CI = -1.68 to 6.01, Cohen's d = 0.45) was moderate in favour of the intervention group, and at 12-month follow-up, semantic verbal fluency gains for the intervention group were detected ({beta} = 2.78, 95% CI = -0.92 to 6.49, Cohen's d = 0.70), though with wide confidence intervals. Conclusions: Despite some small effects observed in memory and verbal fluency, cognitive training did not yield statistically significant improvements. The observed low adherence and limited benefits on mild cognitive deficits in mostly middle-aged individuals with MetS are likely associated with the self-administered and short-term nature of the computerized intervention. This highlights the need for more intensive and clinician-delivered approaches to enhance engagement. Registry: ClinicalTrials.gov, TRN: NCT05658354, Registration date: 08 December 2022. Keywords: Metabolic syndrome, cognitive deficits, cognitive training, computerized, adults

15.
arXiv (CS.AI) 2026-06-25

Surrogate models for Rock-Fluid Interaction: A Grid-Size-Invariant Approach

arXiv:2602.22188v2 Announce Type: replace-cross Abstract: Modelling rock-fluid interaction requires solving a set of partial differential equations (PDEs) to predict the flow behaviour and the reactions of the fluid with the rock on the interfaces. Conventional high-fidelity numerical models require a high resolution to obtain reliable results, resulting in huge computational expense. This restricts the applicability of these models for multi-query problems, such as uncertainty quantification and optimisation, which require running numerous scenarios. As a cheaper alternative to high-fidelity models, this work develops eight surrogate models for predicting the fluid flow in porous media. Four of these are reduced-order models (ROM) based on one neural network for compression and another for prediction. The other four are single neural networks with the property of grid-size invariance; a term which we use to refer to image-to-image models that are capable of inferring on computational domains that are larger than those used during training. In addition to the novel grid-size-invariant framework for surrogate models, we compare the predictive performance of UNet and UNet++ architectures, and demonstrate that UNet++ outperforms UNet for surrogate models. Furthermore, we show that the grid-size-invariant approach is a reliable way to reduce memory consumption during training, resulting in good correlation between predicted and ground-truth values and outperforming the ROMs analysed. The application analysed is particularly challenging because fluid-induced rock dissolution results in a non-static solid field and, consequently, it cannot be used to help in adjustments of the future prediction.

16.
arXiv (CS.CV) 2026-06-18

Quantile Transfer for Reliable Operating Point Selection in Visual Place Recognition

Visual Place Recognition (VPR) is a key component for localisation in Global Navigation Satellite System (GNSS)-denied environments, but its performance critically depends on selecting an image matching threshold (operating point) that balances precision and recall. Thresholds are typically hand-tuned offline for a specific environment and fixed during deployment, leading to degraded performance under environmental change. We propose a method that automatically selects the operating point of a VPR system to maximise recall at 100% precision. The method uses a small calibration traversal with known correspondences and transfers thresholds to deployment via quantile normalisation of similarity score distributions. This quantile transfer ensures that thresholds remain stable across calibration sizes and query subsets. Experiments with seven state-of-the-art VPR techniques across five benchmark datasets demonstrate that our proposed approach consistently outperforms existing baselines, enabling the underlying VPR technique to operate at 100% precision in approximately twice as many deployment scenarios (median improvement), while retrieving up to 29% more correct matches at that precision. The method eliminates manual tuning by adapting to new environments and generalising across operating conditions. Our code is available at https://github.com/DhyeyR-007/Quantile-Transfer-for-Reliable-VPR.

17.
arXiv (CS.AI) 2026-06-19

ProMUSE: Progressive Multi-modal Uncertainty-guided Staged Evidential Alzheimer Disease Classification

arXiv:2606.19371v1 Announce Type: cross Abstract: Alzheimer's disease (AD) is a fatal disorder that destroys memory and cognitive skills in the elderly population. Most treatments for AD are effective in the early stage, leading to an increasing demand for early AD diagnosis. AD diagnosis increasingly relies on multimodal data such as clinical assessments, structural Magnetic Resonance Imaging (MRI), and Positron Emission Tomography (PET) imaging. However, MRI and PET acquisition remain costly and not universally accessible, making full-modality inference impractical in real-world clinical workflows. We propose ProMUSE, a Progressive Multi-modal Uncertainty Guided Staged Evidential Network that adaptively determines when additional modalities are necessary, helping reduce the overall cost of data acquisition while maintaining accuracy. ProMUSE first performs evidential classification using low-cost clinical data and quantifies uncertainty via a Dirichlet-based subjective logic model. When uncertainty exceeds a learned threshold, ProMUSE progressively incorporates MRI or PET features, fusing modality-wise belief and uncertainty through Dempster-Shafer theory to obtain a calibrated multimodal prediction. This staged acquisition strategy enables accurate diagnosis while minimizing reliance on expensive imaging. Experiments on ADNI, AIBL, and OASIS across CN-AD, CN-MCI, and MCI-AD tasks demonstrate that ProMUSE achieves competitive or superior accuracy compared to full-modality baselines while reducing MRI/PET usage by 50-90%, yielding substantial cost savings. These results highlight ProMUSE as a practical, uncertainty-aware, and resource-efficient solution for real-world AD screening.

18.
arXiv (CS.LG) 2026-06-16

ML Inference Scheduling with Predictable Latency

arXiv:2512.18725v3 Announce Type: replace Abstract: Machine learning (ML) inference serving systems can schedule requests to improve GPU utilization and to meet service level objectives (SLOs) or deadlines. However, improving GPU utilization may compromise latency-sensitive scheduling, as concurrent tasks contend for GPU resources and thereby introduce interference. Given that interference effects introduce unpredictability in scheduling, neglecting them may compromise SLO or deadline satisfaction. Nevertheless, existing interference prediction approaches remain limited in several respects, which may restrict their usefulness for scheduling. First, they are often coarse-grained, which ignores runtime co-location dynamics and thus restricts their accuracy in interference prediction. Second, they tend to use a static prediction model, which may not effectively cope with different workload characteristics. In this paper, we evaluate the potential limitations of existing interference prediction approaches, finding that coarse-grained methods can lead to noticeable deviations in prediction accuracy and that static models degrade considerably under changing workloads.

19.
Nature (Science) 2026-06-23

How should I respond to race-based exclusion in my lab?

作者:

A researcher in Europe feels left out of their team and held to different standards from their colleagues. How can they challenge exclusion without risking their position? A researcher in Europe feels left out of their team and held to different standards from their colleagues. How can they challenge exclusion without risking their position?

20.
arXiv (CS.CL) 2026-06-11

Reassessing High-Performing LLMs on Polish Medical Exams: True Competence or Bias-Driven Performance?

Large language models (LLMs) in medicine are mainly evaluated using multiple-choice question answering (MCQA), which can overestimate real clinical ability due to guessing strategies and answer biases. To address these limitations, we introduce an expanded and more challenging benchmark based on Polish medical exams, adding over 15,000 questions, two new domains, and four structural modifications that reduce MCQA-specific artifacts and better test reasoning. We evaluate 21 LLMs and show that evaluation design strongly affects results. Under our harder setup, the best model (Qwen3.5-122B) drops by 28.4 and 31 pp on English and Polish exams, respectively. Despite low evidence of data contamination, standard MCQA scores do not reliably reflect true medical competence. To facilitate further research, we make our benchmark publicly available.

21.
arXiv (CS.CL) 2026-06-11

Pass@K Policy Optimization: Solving Harder Reinforcement Learning Problems

Reinforcement Learning (RL) algorithms sample multiple n>1 solution attempts for each problem and reward them independently. This optimizes for pass@1 performance and prioritizes the strength of isolated samples at the expense of the diversity and collective utility of sets of samples. This under-utilizes the sampling capacity, limiting exploration and eventual improvement on harder examples. As a fix, we propose Pass-at-k Policy Optimization (PKPO), a transformation on the final rewards which leads to direct optimization of pass@k performance, thus optimizing for sets of samples that maximize reward when considered jointly. Our contribution is to derive novel low variance unbiased estimators for pass@k and its gradient, in both the binary and continuous reward settings. We show optimization with our estimators reduces to standard RL with rewards that have been jointly transformed by a stable and efficient transformation function. While previous efforts are restricted to k=n, ours is the first to enable robust optimization of pass@k for any arbitrary k

22.
arXiv (CS.LG) 2026-06-16

Remember, Don't Re-read: Stateful ReAct Agents for Token-Efficient Autonomous Experimentation

arXiv:2606.14945v1 Announce Type: new Abstract: The autoresearch pattern enables autonomous experimentation by having a large language model (LLM) iteratively modify code to optimize a target metric. Its stateless design, however, reconstructs experimental context from scratch at every iteration, incurring $O(n)$ token cost per iteration and $O(n^{2})$ total. This work reformulates the pattern as a stateful ReAct agent using LangGraph, where typed persistent state carries experimental history across iterations via a tool-calling interface. Two benchmarks are evaluated: hyperparameter tuning (15 iterations, small per-iteration observations) and code performance optimization (40 iterations, large per-iteration observations containing full source code and benchmark results). On hyperparameter tuning, the stateful agent consumes 90\% fewer tokens (2{,}492 vs.\ 24{,}465). On code optimization, the stateful agent consumes 52\% fewer tokens (627K vs.\ 1{,}275K) while achieving comparable optimization quality on both tasks. The token reduction is structural: the stateless agent re-reads the full history at $O(n)$ cost per iteration, while the stateful agent operates within a fixed-size conversation window at $O(1)$ cost. This paper describes the architecture in sufficient detail for practitioners to implement a stateful autoresearch agent for their own workflows.

23.
arXiv (CS.CV) 2026-06-17

Bridging Modality Disconnect in Self-Reflection via Closed-Loop Visually Grounded Verification

In the era of Vision-Language Models (VLMs), enhancing multimodal reasoning capabilities remains a critical challenge, particularly in handling ambiguous or complex visual inputs, where initial inferences often lead to hallucinations or logic errors. Existing VLMs often produce plausible yet ungrounded answers, and even when prompted to "reflect", their corrections may remain detached from the image evidence. To address this, we propose the MIRROR framework for Multimodal Iterative Reasoning via Reflection On visual Regions. By embedding visual reflection as a core mechanism, MIRROR is formulated as a closed-loop process comprising draft, critique, region-based verification, and revision, which are repeated until the output is visually grounded. To facilitate training of this model, we construct **ReflectV**, a visual reflective dataset for multi-turn supervision that explicitly contains reflection triggers, region-based verification actions, and answer revision grounded in visual evidence. Experiments on both general vision-language benchmarks and representative vision-language reasoning benchmarks show that MIRROR improves correctness and reduces visual hallucinations, demonstrating the value of training reflection as an evidence-seeking, region-aware verification process rather than a purely textual revision step.

24.
arXiv (CS.AI) 2026-06-12

EWAM: An Enhanced World Action Model for Closed-Loop Online Adaptation in Embodied Intelligence

arXiv:2606.12690v1 Announce Type: cross Abstract: In this paper, we propose the Enhanced World Action Model (EWAM), a closed-loop online adaptation architecture built upon a pretrained and fully frozen Cosmos3 backbone network. Evaluated entirely under a zero-shot task protocol, EWAM is centrally focused on reducing the amount of additional deployment data required to adapt to new task layouts. Notably, no extra task-specific demonstration sets were introduced in any of the evaluations, and no fine-tuning was performed on the backbone network. Its performance gains stem entirely from an inference-time co-reasoning mechanism composed of four inserted lightweight neural layers: the Neural Experience Memory Layer located in the intermediate layers of the Diffusion Transformer (DiT) provides task-relevant execution context; the Neural Anomaly Detection Layer after the state prediction head monitors the divergence between predicted and actual states in real time; the Neural Policy Routing Layer dynamically selects direct execution, conservative replanning, or rollback recovery based on the anomaly severity; and the Neural Action Correction Layer refines the generated action chunks using execution diagnostics. Unlike naive feature fusion, the memory, anomaly detection, and correction modules are deeply integrated into the Cosmos3 forward path in a differentiable manner, with only the final routing decision being a discrete supervised one.

25.
arXiv (CS.LG) 2026-06-15

CANN-EUCLID: unsupervised constitutive artificial neural network model discovery from full-field data

arXiv:2606.14565v1 Announce Type: cross Abstract: Constitutive artificial neural networks (CANNs) provide interpretable material model discovery, but have so far been used in stress-supervised settings based on apparent stress-strain data from homogeneous tests. Because each test samples only a narrow loading path and provides homogenized rather than local stress information, robust discovery typically requires multiple loading modes to constrain the multidimensional response. This is challenging for soft biological tissues, where repeated testing, damage, and sample variability limit reliable information from a single specimen. Here, we combine CANNs with the stress-unsupervised full-field discovery framework EUCLID to identify sparse hyperelastic laws directly from displacement fields and reaction forces in one heterogeneity-inducing loading case. CANN-EUCLID minimizes equilibrium imbalance with sparsity-promoting regularization selecting compact active terms, without local stress measurements or a prescribed law. We evaluate the approach on isotropic and anisotropic benchmarks with prescribed ground-truth laws. When the ground truth is representable by the chosen CANN basis, our method recovers the correct terms with near-exact accuracy, including exponential terms with embedded parameters. When it is not contained in the basis, the method retains shared terms and approximates missing contributions using available basis functions. Generalization depends strongly on sampled deformation states: exponential strain-stiffening terms can be recovered accurately when sufficiently probed, but can produce large extrapolation errors when the stiffening regime lies outside the sampled domain. Forward FE validation simulations show that the discovered behavior accurately replicates the ground truth. These results establish stress-unsupervised CANN discovery as a promising framework for interpretable full-field constitutive model identification.