01.
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。
02.
arXiv (CS.AI)
2026-06-12
Functional Cache Grafting: Robust and Rapid Code-Policy Synthesis for Embodied Agents
作者:
arXiv:2606.13097v1 Announce Type: cross Abstract: Code-writing large language models (CodeLLMs) generate executable code policies for embodied agents by translating natural language goals and environmental constraints into structured control programs. However, policy generation in open-domain embodied environments suffers from two fundamental limitations: (i) delayed decoding caused by repetitive prefill computation over long prompts, and (ii) limited robustness due to fully generative decoding, which often produces API mismatches, missing safety guards, and unstable control logic. To address these limitations, we present FCGraft, a Functional Cache Grafting framework. FCGraft maintains a library of function-level validated code skeletons and their associated prompt-level Transformer key-value (KV) caches, and synthesizes new policies by retrieving relevant functions and grafting their KV caches when a new task is provided. Given retrieved function caches, FCGraft performs cache grafting via stitching, which composes cached function segments into a composite policy, and patching, which locally adapts only the necessary code regions to satisfy task-specific parameters and constraints with minimal additional decoding. By eliminating redundant prefill computation, this approach reduces generation latency, while reusing validated control structures improves robustness over prompt-level caching methods RAGCache, achieving 18.31% higher task success rate and 2.3x faster policy synthesis.
03.
medRxiv (Medicine)
2026-06-16
Adherence to Red Reflex and Vision Screening Recommendations: A Deep Dive into Primary Care Implementation Gaps
作者:
…
Introduction: Early childhood vision screening is critical for detecting amblyopia and other vision-threatening conditions. Despite screening recommendations during well-child visits, rates remain low. Red reflex assessment is recommended to identify serious ocular pathology, yet its use in primary care is not well described. We examined rates and drivers of vision screening in pediatric primary care. Methods: We conducted a retrospective review of electronic health records for children 3 to 5 years attending well-child visits in 2022 in one of three representative primary care clinics within a university health system. Outcomes were documented red reflex and functional vision tests. We evaluated associations with patient demographics and clinic site using multivariable logistic regression Results: Among 1,003 visits, 21.1% (n=212) had a documented red reflex assessment, and 60.8% (n=610) a functional vision test. Younger children (ages 3 and 4 vs. 5 years) had higher odds of red reflex assessment [adjusted odds ratio (aOR) 9.00 and 8.64], and lower odds of a functional vision (aOR 0.47 and 0.59) test. Females had higher odds of red reflex assessment (aOR 1.53). Other/Multiracial children had lower odds of red reflex assessment than Non-Hispanic White children (aOR 0.48). Screening rates varied significantly by clinic site Conclusions: Visual function and red reflex assessment are inconsistently performed in pediatric primary care, with particularly low rates of red reflex documentation. Screening rates varied between clinics and were affected by age. These findings highlight missed opportunities for early detection of vision-threatening conditions and identify targets for improving adherence to pediatric vision screening recommendations
04.
arXiv (CS.CL)
2026-06-17
RubricsTree: Scalable and Evolving Open-Ended Evaluation of Personal Health Agents across Health Memory and Medical Skills
作者:
…
The LLM-empowered personal health agents with user health (sensor) metrics have offered a promising pathway to alleviate global disparities in healthcare access. However, large-scale clinical deployment remains constrained by an open-ended evaluation bottleneck: physician annotation is reliable but costly and unscalable, while LLM-as-a-judge evaluators are scalable but subjective, inconsistent, and sometimes clinically misaligned. We introduce RubricsTree, a scalable evaluation framework with an expert-aligned hierarchical taxonomy of over 100 atomic, clinically-verifiable Boolean rubrics, evolving from the insights of 4,000 real user queries through an iterative human-in-the-loop curation protocol with an expertise panel led by an experienced physician. A context-aware adaptive router activates only the relevant auto-weighted rubric subset per query, providing the throughput needed for scalable evaluation with expert-aligned quality. Through a systematic meta-evaluation, we show that RubricsTree (i) substantially exceeds a strong large-scale evaluation baseline in expert alignment on challenging open-ended queries; (ii) reliably penalizes contextually degraded responses; and (iii) when used as structured instructions, text feedback, or training rewards for performance optimization, yields up to ~66% relative gains on HealthBench for Gemini, GPT, and Qwen model families. RubricsTree thus provides a scalable, auditable, and evolving evaluation infrastructure required for the continuous optimization of product-level personal healthcare AI.
05.
arXiv (CS.CL)
2026-06-15
Right or Wrong, Models Comply: Directional Blindness in LLM Moral Judgment
作者:
As language models take integrated roles across many domains, the response of LLMs to user pushback becomes a critical alignment property. Yet many existing evaluations treat compliance as unidirectional, measuring whether models resist pressure but not whether they resist it selectively. We introduce Compliance Asymmetry (A = BCR/HCR), a bidirectional diagnostic that compares beneficial output change under helpful nudges with harmful change under misleading nudges. Across 9 models and 972,000 nudge-condition responses, we find that this selectivity differs in factual and moral judgments: models follow helpful nudges more than harmful ones on factual questions (A = 1.58), but follow both directions at nearly identical rates on moral questions (A = 1.04). This phenomenon persists across model families, capability levels, and nudging types. Interestingly, we also find that chain-of-thought prompting amplifies helpful and harmful compliance together, while identity-based prompting suppresses both by nearly identical margins. These results identify direction-blind moral compliance as a distinct failure mode in current LLMs and suggest that alignment should target directionally calibrated updating rather than lower compliance alone.
06.
arXiv (CS.CL)
2026-06-11
Pre-AF 13: An Interpretable Atrial Fibrillation Risk Score Mined from Discharge Reports
作者:
…
Background. Atrial fibrillation (AF) is the most prevalent cardiac arrhythmia and a major determinant of prognosis. Established AF risk scores rely on factors (older age, hypertension) nearly ubiquitous among patients with cardiovascular disease (CVD), offering limited stratification in this high-risk group. Most target long-term (5-10 year) rather than medium-term prediction. We developed interpretable ML models predicting AF risk over a 24-month and entire follow-up horizon in CVD patients using routinely collected hospital data. Methods. Single-center retrospective study of electronic health records from the National Research Cardiology Center (Russia) for patients aged >=18 with CVD but without pre-existing AF, hospitalized more than once between January 2012 and May 2019. A custom NLP pipeline transformed unstructured discharge reports into 73 structured features, combining a rule-based parser with transformer-based NER. Using LightAutoML we built a full model (73 features), a simple model (reduced subset), and a linear model for a bedside risk score. Performance was assessed by ROC AUC, compared with CHARGE-AF, C2HEST, MHS, and HAVOC, and interpreted via SHAP. Results. Of 80,576 records from 45,000 patients, 17,562 met inclusion criteria; 1,438 (8.19%) developed AF. The full model reached ROC AUC 0.735 (24-month) and 0.696 (entire follow-up); the simple model was nearly identical (0.725, 0.696). All non-linear models outperformed the four clinical risk scores (ROC AUC 0.53-0.64). The simple model uses 13 features and is named Pre-AF 13. SHAP identified age and left atrial volume as dominant predictors. A linear risk score (Pre-AF 9) stratified observed 24-month AF incidence from ~7% to 36%. Conclusion. Interpretable ML models built from routinely collected EHR data identify high-AF-risk CVD patients, outperforming established clinical risk scores.
07.
arXiv (quant-ph)
2026-06-15
Scaling native entanglement generation in layered semiconductors with quasi-phase matching
作者:
…
arXiv:2606.14553v1 Announce Type: new Abstract: Efficient generation of entangled photons typically relies on spontaneous parametric down-conversion (SPDC) in phase-matched macroscopic nonlinear media. However, generating entanglement under phase-matching constraints requires additional bulk optics or interferometers. In contrast, ultrathin van der Waals semiconductors - such as transition metal dichalcogenides (TMDs) - exhibit strong enough optical nonlinearities for SPDC to be observed from subwavelength-thick media, thereby bypassing conventional phase-matching constraints. In this microscopic domain, the intrinsic crystal symmetry governs the nonlinear optical response, enabling the native generation of polarization-entangled photon pairs. However, generating these states efficiently has been fundamentally restricted by the material's coherence length ($L_c$), which limits the attainable conversion efficiency. Here, we investigate periodically-poled TMDs (PPTMDs) designed to scale up this interaction via quasi-phase matching. We demonstrate that mechanically flipping the sign of the nonlinearity at precise intervals of $L_c$ introduces quasi-phase matching, that scales the pair-production rate while preserving the pristine, symmetry-generated polarization entanglement, with fidelities exceeding 99%. Backed by a rigorous theoretical model, our work clarifies the interplay between crystal symmetry and propagation effects in thin nonlinear media, providing a new avenue for engineering quantum light in nanophotonic systems.
08.
Nature (Science)
2026-06-10
The Amazon can be saved — with concerted action inside and outside Brazil
作者:
未知作者
As deforestation in the Amazon falls, fresh evidence shows that the rainforest can withstand global warming, but only if there is a worldwide effort to stop cutting it down. As deforestation in the Amazon falls, fresh evidence shows that the rainforest can withstand global warming, but only if there is a worldwide effort to stop cutting it down.
09.
Nature (Science)
2026-06-17
Navigating a crowded developing brain leaves neurons with broken DNA
作者:
As neurons migrate to their final destinations in the forming brain, their DNA gets damaged. The brain has evolved a fix, but there can be lasting consequences if repair fails. As neurons migrate to their final destinations in the forming brain, their DNA gets damaged. The brain has evolved a fix, but there can be lasting consequences if repair fails.
10.
arXiv (CS.LG)
2026-06-16
Biarchetype analysis for univariate functional data. An application to macroeconomic financial time series
作者:
arXiv:2606.15881v1 Announce Type: cross Abstract: We introduce biarchetype analysis for the first time in the context of univariate functional data. This unsupervised methodology extends archetype analysis by simultaneously identifying archetypal structures across both the cases (countries, in our application) and the temporal argument. Both cases and time points are expressed as mixtures of biarchetypes, yielding a concise and highly interpretable representation of complex functional observations. Although biarchetype analysis is not intended as a clustering technique, it offers superior interpretability compared with biclustering approaches, as it is based on extreme, representative patterns rather than average centroids, thereby enhancing human comprehension. We apply the proposed method to 10-year government bond yields of European countries over the period 2001-2025. The results identify three distinct time regimes (the pre-crisis period, the euro-area sovereign debt crisis, and the post-crisis period), and reveal Germany, Greece, and Hungary as country archetypes.
11.
Nature (Science)
2026-06-10
Daily briefing: Ancient ground squirrels ate like ‘zombies of the Pleistocene’
作者:
Evidence from fossilized poo reveals the diverse diet of ancient ground squirrels. Plus, the science behind the peptide craze and our innate tendency to wander anticlockwise. Evidence from fossilized poo reveals the diverse diet of ancient ground squirrels. Plus, the science behind the peptide craze and our innate tendency to wander anticlockwise.
12.
arXiv (CS.AI)
2026-06-12
Structured vs. Unstructured Pruning: An Exponential Gap
作者:
arXiv:2603.02234v3 Announce Type: replace-cross Abstract: The Strong Lottery Ticket Hypothesis (SLTH) states that large, randomly initialized neural networks contain sparse subnetworks capable of approximating a target function at initialization without training, suggesting that pruning alone is sufficient. Pruning methods are typically classified as unstructured, where individual weights can be removed from the network, and structured, where parameters are removed according to specific patterns, as in neuron pruning. Existing theoretical results supporting the SLTH rely almost exclusively on unstructured pruning, showing that logarithmic overparameterization suffices to approximate simple target networks. In contrast, neuron pruning has received limited theoretical attention, despite its practical appeal for direct hardware speedups. In this work, we consider the problem of approximating a single bias-free ReLU neuron by pruning hidden units of a randomly initialized two-layer ReLU network, effectively isolating the intrinsic limitations of neuron pruning. We show that achieving an $\varepsilon$-approximation requires a starting network size of $\Omega(1/\varepsilon)$ for neuron pruning, whereas weight pruning succeeds with only $O(\log(1/\varepsilon))$ hidden units, revealing an exponential separation between the two approaches.
13.
arXiv (CS.LG)
2026-06-16
Analytic Torsion and Spectral Gap Capture Persistent-Laplacian Performance
作者:
arXiv:2606.16990v1 Announce Type: new Abstract: While persistent Laplacians (PL) offer a richer geometric representation of data than persistent homology, utilizing their full eigenspectrum for learning tasks is often hampered by high dimensionality and the ``varying length'' problem across different filtration scales. We propose a compact spectral representation that distills the persistent Laplacian into three mathematically grounded invariants: Betti numbers, the spectral gap, and analytic torsion. Across benchmark datasets including MNIST, QM-3D, and SKEMPI WT, we demonstrate that this reduced feature space captures the essential predictive signal of the full spectrum, and in some cases outperforms it, while significantly reducing computational overhead and preventing the noise introduced by higher-frequency eigenvalues. Our results suggest that these invariants provide a principled, fixed-length interface between spectral geometry and topological learning.
14.
arXiv (CS.CL)
2026-06-19
Where Does Social Reasoning Come From? Capability Provenance in Language Models
作者:
We use training-data attribution as an interpretable tool for capability discovery, mapping which regions of the pretraining corpus support social-reasoning versus STEM-reasoning in OLMo3-7B. Training-data attribution measures how strongly each training document influences a model's predictions on a benchmark, but document-level scores are too noisy to identify which corpus regions support which capabilities, and prior work has emphasized factual knowledge rather than reasoning. We compute gradient-based attribution (TrackStar via Bergson) over a working set drawn from the de-duplicated Dolma3 mix, aggregate influence across WebOrganizer's 24-format x 24-topic taxonomy (576 bins), and contrast benchmark pairs in a 2x2 design that varies domain (social vs. STEM) and capability type (reasoning vs. knowledge): SocialIQA and MMLU Social Sciences against ARC-Challenge and MMLU STEM. Social and STEM reasoning draw on qualitatively distinct corpus regions, and the contrast is sharper at the reasoning level than at the knowledge level. Targeted machine unlearning provides partial causal validation: forgetting high-attribution topic bins (e.g., Literature for SocialIQA) degrades the aligned benchmark more than within-bin random baselines, and we open-source all code, sampling manifests, the bin-level influence matrix, and unlearning checkpoints.
15.
arXiv (CS.CV)
2026-06-11
From Prompts to Tokens: Internalizing Causal Supervision in Vision-Language Model for Multi-Image Causal Reasoning
作者:
Visual causal reasoning is essential for understanding and intervening in the physical world, requiring identification of causal variables from visual inputs and reasoning over intervention effects. Despite recent progress, large vision–language models (VLMs) remain brittle at such tasks, especially for interventional and counterfactual queries over multi-image inputs. Most existing explorations inject causal knowledge via textual prompts, leaving causal mechanisms external to model execution and limiting reliable control during inference. To address this problem, we propose BridgeVLM, which internalizes visual causal reasoning by inducing a causal graph from multi-image inputs and converting it into structured Causal Tokens executed by RAMP layers injected into the LLM decoder for causal message passing. We further introduce a unified training interface M3S for fine-grained causal supervision from different granularities (local/global level). BridgeVLM achieves 54.4% accuracy on intervention tasks on CausalVLBench (vs. 33.2% with prompt-level supervision), improves results on Causal3D from 43.6% to 49.0%, and substantially improves causal structure learning on CausalVLBench ($F_1$: 33.4% $\rightarrow$ 75.1%).
16.
arXiv (quant-ph)
2026-06-15
Computational regimes in matrix-product-state-based quantum trajectory simulations
作者:
arXiv:2606.13779v1 Announce Type: new Abstract: Efficient simulation of open quantum systems is central to modeling noisy quantum hardware and many-body dynamics. In trajectory-based tensor network methods, cost is often associated with trajectory-level quantities such as entanglement growth or bond dimension. However, the total cost of a fixed-accuracy simulation also depends on statistical sampling, and the interplay between per-trajectory complexity and sampling effort remains poorly understood. Here we introduce a cost-resolved framework for matrix product state (MPS)-based quantum trajectory simulations that decomposes total cost into memory per trajectory, runtime per trajectory, and sampling effort. We show that physically equivalent stochastic unravelings of the same Lindblad dynamics do not necessarily reduce total cost, but instead redistribute cost between trajectory complexity and statistical convergence. This trade-off is quantified by two dimensionless inflation factors: a bond dimension inflation $\alpha$ and a sampling inflation $\kappa$, which together determine the preferred unraveling under hardware-dependent memory and parallelism constraints. We provide a practical protocol for extracting $(\alpha,\kappa)$ from modest pilot simulations and demonstrate it using benchmarks across multiple noise channels. The resulting decision maps show that the computationally favorable unraveling can change with noise strength, time-step resolution, system size, and available parallelism. These results establish unraveling choice as a hardware-aware simulation design problem rather than an intrinsic optimization of trajectory entanglement alone.
17.
medRxiv (Medicine)
2026-06-23
THE SILENT STRUGGLE: EXPLORING THE EFFECTS OF COMMUNICATION BREAKDOWNS IN HEALTHCARE DELIVERY IN THE NORTHERN REGION OF GHANA
作者:
Abstract Effective health communication is central to patient-centred care and improved health outcomes, particularly in culturally diverse healthcare settings. In clinical and assistive practice, communication breakdowns may negatively affect diagnosis, treatment adherence, and preventive care. A qualitative phenomenological design was employed, utilizing Semi-Structured interviews with purposively sampled twenty patients and healthcare professionals from Tamale Teaching Hospital, Yendi Hospital, and Bimbilla Hospital. The researchers adopted Content Analysis as the tool of analysis for the data. The findings of this study revealed that language discrepancies Poor attitudes of healthcare providers hinderer patient openness and the quality treatment. Logistical issues, such as inadequate medicines and medical supplies, resulted in delayed treatment and additional financial burden on patients and their relatives. Cultural and social factors discourage patients from discussing certain health conditions with healthcare providers, leading to delayed treatment. These hurdles adversely impact on treatment and assistive practice, specifically in culturally diverse environment and preventive care. The study recommends training and capacity-building programs for healthcare providers in cultural competence, fostering effective and ethical health communication between patients and healthcare providers, and recruiting professional interpreters to bridge the linguistics gap between patients and providers. Abstract Effective health communication is central to patient-centered care and improved health outcomes, particularly in culturally diverse healthcare settings. In clinical and assistive practice, communication breakdowns may negatively affect diagnosis, treatment adherence, and preventive care. A qualitative phenomenological design was employed, utilizing semi-structured interviews with twenty purposively sampled patients and healthcare professionals from Tamale Teaching Hospital, Yendi Hospital, and Bimbilla Hospital. The researchers adopted content analysis as the tool of analysis for the data. The findings of this study revealed that language discrepancies Poor attitudes of healthcare providers hinder patient openness and quality treatment. Logistical issues, such as inadequate medicines and medical supplies, resulted in delayed treatment and additional financial burden on patients and their relatives. Cultural and social factors discourage patients from discussing certain health conditions with healthcare providers, leading to delayed treatment. These hurdles adversely impact treatment and assistive practice, specifically in culturally diverse environments and preventive care. The study recommends training and capacity-building programs for healthcare providers in cultural competence, fostering effective and ethical health communication between patients and healthcare providers, and recruiting professional interpreters to bridge the linguistics gap between patients and providers.
18.
medRxiv (Medicine)
2026-06-22
Histologically validated diffusion MRI signatures of neuroinflammation and neurodegeneration in Alzheimer disease
作者:
…
Noninvasive neuroinflammation measurement remains a major barrier for Alzheimer disease (AD) therapeutics. We present generalized diffusion basis spectrum imaging (g-DBSI), a diffusion MRI framework that decomposes the tissue signal into biologically interpretable microstructural compartments. In postmortem Knight ADRC brains, g-DBSI-derived restricted isotropic fraction (RIF) and restricted anisotropic fraction (RAF) mapped cellularity and neurofilament density, while their ratio (RIF/RAF) tracked inflammatory cell density and peri-plaque amyloid-beta with higher specificity and regional consistency than RIF alone. In 112 living Knight ADRC participants stratified by PET amyloid, g-DBSI metrics showed amyloid-dependent trajectories: in low-amyloid individuals, RIF and RAF rose together with amyloid, consistent with early neuropil expansion and glial elaboration, whereas in high-amyloid individuals, RIF/RAF increased, and RAF declined, indicating established neuroinflammatory remodeling and neurofilament loss. CSF proteomics linked RIF/RAF to glia-enriched immune and vascular pathways, supporting g-DBSI as a clinically compatible MRI biomarker of neuroinflammation and neurodegeneration in AD.
19.
arXiv (CS.CL)
2026-06-16
JE-IRT: A Geometric Lens on LLM Abilities through Joint Embedding Item Response Theory
作者:
Standard LLM evaluation practices compress diverse abilities into single scores, obscuring their inherently multidimensional nature. We present JE-IRT, a geometric item-response framework that embeds both LLMs and questions in a shared space. For question embeddings, the direction encodes semantics and the norm encodes difficulty, while correctness on each question is determined by the geometric interaction between the model and question embeddings. This geometry replaces a global ranking of LLMs with topical specialization and enables smooth variation across related questions. Building on this framework, our experimental results reveal that out-of-distribution behavior can be explained through directional alignment, and that larger norms consistently indicate harder questions. Moreover, JE-IRT naturally supports generalization: once the space is learned, new LLMs are added by fitting a single embedding. The learned space further reveals an LLM-internal taxonomy that only partially aligns with human-defined subject categories. We also show that simple linear probes of the embedding space recover cross-subject ability directions, such as an arithmetic axis that highlights quantitatively demanding questions in seemingly distant subjects like virology and global facts. JE-IRT thus establishes a unified and interpretable geometric lens that connects LLM abilities with the structure of questions, offering a distinctive perspective on model evaluation and generalization.
20.
arXiv (math.PR)
2026-06-12
Storage and Transport Capacity Design for a Self-Reliable Two-Node Stochastic Resource System
作者:
arXiv:2606.12707v1 Announce Type: cross Abstract: We study a two-node stochastic resource system operating over a finite horizon. Each node experiences uncertain supply and demand and is equipped with finite storage. The objective is to ensure that resource levels remain within prescribed limits with high probability. To this end, we formulate a chance-constrained capacity-design problem in which resources can be exchanged through a capacity-limited transport link. We characterize the minimum storage required at each node, derive the optimal transport policy, and quantify the trade-off between storage and transport capacities. Our results show the existence of a critical transport-capacity threshold that enables full risk pooling between the nodes. Moreover, this threshold decreases with the operating horizon, implying that full-pooling performance can be achieved with progressively smaller transport capacity over longer horizons.
21.
medRxiv (Medicine)
2026-06-15
Therapeutic efficacy study on shoulder impingement syndrome in swimmers: a network meta-analysis
作者:
Shoulder impingement syndrome (SIS), including subacromial impingement and rotator cuff tendinitis, is commonly caused by repetitive swimming movements and associated shoulder joint dysfunction. Despite numerous available treatment options, no consensus exists on the most effective treatment option. Therefore, this systematic review and network meta-analysis aimed to investigate treatment methods for SIS in swimmers. Using a frequentist framework and Cochrane PICOS principles, we compared SIS treatments, constructed network evidence diagrams, and assessed heterogeneity. A total of 45 studies were included in the qualitative synthesis, and 42 contributed to the network meta-analysis, comprising 1752 participants, 9 treatment categories, and outcome measures. For pain outcomes, some adjunctive interventions combined with exercise showed favorable ranking probabilities, although several estimates were accompanied by wide confidence intervals. For shoulder range-of-motion outcomes, taping, acupuncture, manual therapy, and sport-specific training showed favorable effects in selected comparisons, particularly for external and internal rotation. According to surface under the cumulative ranking curve (SUCRA) rankings, exercise combined with medium-frequency therapy ranked highly for pain reduction, whereas exercise combined with acupuncture or extracorporeal shock wave therapy ranked highly for shoulder flexion. Exercise combined with taping ranked highly for external rotation, and exercise combined with manual therapy ranked highly for internal rotation. However, the interpretation of ranking results should remain cautious because uncertainty and inconsistency were present in some comparisons. Exercise-based rehabilitation appears to remain central to the management of SIS in swimmers. Several adjunctive interventions showed favorable findings for selected outcomes, especially pain relief and shoulder rotational function. However, the available evidence was affected by heterogeneity, inconsistency, and imprecision across some treatment comparisons. More rigorously designed swimmer-specific randomized controlled trials are needed before firm treatment hierarchies can be established. Trial registration: The protocol for this systematic review is registered with PROSPERO (www.crd.york.ac.uk/PROSPERO; registration number: CRD42024498851). The first submission of PROSPERO was on January 15, 2024, and it was revised and updated on March 25, 2026.
22.
arXiv (CS.LG)
2026-06-17
Exposing the Illusion of Fairness: Auditing Vulnerabilities to Distributional Manipulation Attacks
作者:
arXiv:2507.20708v3 Announce Type: replace Abstract: The rapid deployment of AI systems in high-stakes domains, including those classified as high-risk under the The EU AI Act (Regulation (EU) 2024/1689), has intensified the need for reliable compliance auditing. For binary classifiers, regulatory risk assessment often relies on global fairness metrics such as the Disparate Impact ratio, widely used to evaluate potential discrimination. In typical auditing settings, the auditee provides a subset of its dataset to an auditor, while a supervisory authority may verify whether this subset is representative of the full underlying distribution. In this work, we investigate to what extent a malicious auditee can construct a fairness-compliant yet representative-looking sample from a non-compliant original distribution, thereby creating an illusion of fairness. We formalize this problem as a constrained distributional projection task and introduce mathematically grounded manipulation strategies based on entropic and optimal transport projections. These constructions characterize the minimal distributional shift required to satisfy fairness constraints. To counter such attacks, we formalize representativeness through distributional distance based statistical tests and systematically evaluate their ability to detect manipulated samples. Our analysis highlights the conditions under which fairness manipulation can remain statistically undetected and provides practical guidelines for strengthening supervisory verification. We validate our theoretical findings through experiments on standard tabular datasets for bias detection. Code is publicly available at https://github.com/ValentinLafargue/Inspection.
23.
arXiv (CS.CL)
2026-06-15
The Coin Flip Judge? Reliability and Bias in LLM-as-a-Judge Evaluation
作者:
LLM-as-a-Judge is now widely used to rank model outputs, train reward models, and populate public leaderboards, but its run-to-run reliability remains under-characterized. We study repeated identical evaluations on 29 tasks spanning 10 categories using two OpenAI judge models (GPT-4o-mini and GPT-4.1-mini), with 50 pairwise trials and 50 pointwise trials per question, supplemented by temperature and prompt-sensitivity ablations. Across judges, pairwise preferences flip on average 13.6% of the time, with 28% of questions exceeding a 20% flip rate and one question reaching 56%. GPT-4o-mini also exhibits a significant first-position bias (72% A-majority, p = 0.024). At the same time, mean pointwise score gaps are small (0.19–0.36 on a 10-point scale) and not statistically significant in aggregate, producing a pairwise–pointwise gap: judges frequently choose a winner even when their own scalar scores provide little evidence of a meaningful quality difference. Beyond within-judge instability, cross-judge agreement is only 76% ($\kappa = 0.51$), semantically equivalent prompt templates change majority outcomes in 25% of tested cases, and deterministic decoding reduces but does not eliminate inconsistency. A reliability curve analysis shows that, in our dataset, 11 repeated trials are needed for a majority vote to recover the 50-trial reference verdict with 95% probability on average, rising to 15 for high-variance questions. These findings suggest that single-trial LLM judging is often too noisy for high-stakes evaluation, and that multi-trial aggregation, position randomization, and explicit uncertainty reporting should be standard practice. Because both judges are from a single provider, cross-provider replication remains an important next step.
24.
arXiv (CS.AI)
2026-06-19
Calibration Without Comprehension: Diagnosing the Limits of Fine-Tuning LLMs for Vulnerability Detection in Systems Software
作者:
arXiv:2606.20502v1 Announce Type: cross Abstract: Whether LLMs scoring well on vulnerability benchmarks genuinely reason about security or merely pattern-match on contaminated data remains unresolved. We present CWE-Trace, a framework for LLM vulnerability detection built from 834 manually curated Linux kernel samples spanning 74 CWEs. The framework enforces a strict temporal split (pre-2025 historical set / post-cutoff leakage-free set), preserves context-aware vulnerable–patched pairs, and introduces two diagnostic metrics: the Directional Failure Index (DFI) and Hierarchical Distance and Direction (HDD). We evaluate eight vanilla LLMs and 15 LoRA fine-tuned variants across non-targeted detection, targeted detection, and CWE classification. Our analysis yields two key results. First, data contamination provides no measurable advantage. Function-level analysis shows that 84% of nominally contaminated samples carry no usable memorization signal: vulnerable functions are absent or cross-mapped across datasets, and ~31% of contaminated samples carry CWE misclassification. Second, backbone directional priors dominate fine-tuning. Models exhibit stable, systematic failure modes (DFI ranging from -85.5 to +94.8 pp) that persist from historical to post-cutoff data and resist correction. Fine-tuning shifts the output threshold without changing the decision policy. This is calibration without comprehension: output distributions adapt to training data while the underlying security reasoning remains absent. The weakest backbone at binary detection (DeepSeek-R1) gains the most in coarse CWE classification, revealing that detection and understanding are decoupled capabilities. The best detection score reaches only 52.1% (+2.1 pp above chance); exact CWE ranking remains below 1.3% Top-1 accuracy, confirming that current LLMs lack reliable security reasoning for systems software, regardless of fine-tuning strategy.
25.
arXiv (CS.LG)
2026-06-15
Free Heavy-Tailed Lunch for Muon: A Theoretical Justification of Empirical Success
作者:
arXiv:2606.14560v1 Announce Type: cross Abstract: Non-Euclidean optimisation methods with matrix-valued updates, such as Muon and Scion, have recently shown strong empirical performance for training Transformer models, yet their theoretical advantages over Euclidean methods remain poorly understood. We address this gap in the heavy-tailed non-convex regime, where stochastic gradients have bounded $p$-th central moments, $p \in (1,2]$. We show that certain non-Euclidean methods achieve optimal sample complexity under stronger stationarity measures, while Euclidean methods incur additional dimension-dependent costs. As a consequence, for $m \times n$ matrices, Muon finds an $\varepsilon$-stationary point in nuclear norm within $\mathcal{O}\left(\min\{m, n\} \frac{\Delta_1 L}{\varepsilon^2} \left(\frac \sigma \varepsilon \right)^{\frac p {p-1}}\right)$ samples, absorbing heavy-tailed noise without extra dimension dependence, unlike Euclidean methods. We further prove this sample complexity, including its dimension dependence, is optimal for all first-order methods under nuclear-norm stationarity. Experiments on large language models support our theory. Surprisingly, our results suggest that other Schatten geometries beyond the spectral geometry of Muon can perform competitively in certain settings.