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.CV) 2026-06-16

An Open-Source Monitoring Framework for Data Exploration and Progress Tracking in Multi-Center Radiology Studies

Multi-center studies are crucial for advancing medical and radiological research. Data exploration, collaboration discovery, and study progress monitoring are essential for maximizing their potential. However, in practice these processes often rely on manual communication and shared tables, which quickly become outdated and hinder efficient coordination in large distributed studies. This highlights the need for dedicated monitoring solutions that provide transparent and up-to-date insights into study progress. We propose a lightweight, open-source monitoring architecture for multi-center studies based on the widely used Grafana-Prometheus stack. The framework collects aggregated monitoring metrics from distributed study sites and visualizes them through configurable dashboards. As a real-world deployment example, the framework is integrated into the medical imaging platform Kaapana and evaluated within a large multi-center research network. By deploying our solution within the Germany-wide RACOON consortium, we demonstrate its ability to enable privacy-preserving data exploration and study progress monitoring across all 38 German university clinics. The monitoring framework supports transparent coordination of distributed research activities and can facilitate more efficient management of large-scale multi-center studies. The source code and Kaapana integration are publicly available at https://github.com/MIC-DKFZ/study-monitoring-kaapana.

02.
arXiv (CS.CV) 2026-06-17

Fluently Lying: Adversarial Robustness Can Be Substrate-Dependent

The primary tools used to monitor and defend object detectors under adversarial attack assume that when accuracy degrades, detection count drops in tandem. This coupling was assumed, not measured. We report a counterexample observed on a single model: under standard PGD, EMS-YOLO, a spiking neural network (SNN) object detector, retains more than 70% of its detections while mAP collapses from 0.528 to 0.042. We term this count-preserving accuracy collapse Quality Corruption (QC), to distinguish it from the suppression that dominates untargeted evaluation. Across four SNN architectures and two threat models (l-infinity and l-2), QC appears only in one of the four detectors tested (EMS-YOLO). On this model, all five standard defense components fail to detect or mitigate QC, suggesting the defense ecosystem may rely on a shared assumption calibrated on a single substrate. These results provide, to our knowledge, the first evidence that adversarial failure modes can be substrate-dependent.

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

A Multimodal Approach to Alzheimer's Diagnosis: Geometric Insights from Cube Copying and Cognitive Assessments

arXiv:2512.16184v2 Announce Type: replace Abstract: Early and accessible detection of Alzheimer's disease (AD) remains a critical clinical challenge, and cube-copying tasks offer a simple yet informative assessment of visuospatial function. This work proposes a multimodal framework that converts hand-drawn cube sketches into graph-structured representations capturing geometric and topological properties, and integrates these features with demographic information and neuropsychological test (NPT) scores for AD classification. Cube drawings are modeled as graphs with node features encoding spatial coordinates, local graphlet-based topology, and angular geometry, which are processed using graph neural networks and fused with age, education, and NPT features in a late-fusion model. Experimental results show that graph-based representations provide a strong unimodal baseline and substantially outperform pixel-based convolutional models, while multimodal integration further improves balanced classification performance and discriminative ability. SHAP-based interpretability analysis identifies specific graphlet motifs associated with corner integrity and edge continuity as key predictors, closely aligning with clinical observations of distorted cube drawings in AD. Together, these findings establish graph-based analysis of cube-copying behavior as an interpretable, non-invasive, and scalable framework for Alzheimer's disease screening.

04.
arXiv (CS.LG) 2026-06-18

Do as the Romans Do: Learning Universal Behaviors from Heterogeneous Agents

arXiv:2606.18537v1 Announce Type: new Abstract: Humans often acquire new skills by observing others, since observed behaviors implicitly reveal how to act in an environment. However, observations drawn from a heterogeneous population introduce conflicting behavioral signals, making it difficult to determine which behaviors are worth imitating. We address this challenge with General Reward Inference and Disentanglement (GRID), a social learning method that extracts universally useful behaviors from a heterogeneous population of demonstrators pursuing different goals. GRID decomposes per-agent reward functions into a general reward, capturing behaviors shared across all agents, and specific rewards, capturing individual preferences and objectives. Training exclusively on the general reward provides a new paradigm of generalist pretraining. It yields a generalist agent that internalizes universal environmental competencies, such as safety and basic task proficiency, without the mode-averaging bias that afflicts standard learning from demonstration techniques. This generalist serves as a superior prior for fine-tuning to downstream tasks, including preferences unseen during training. Experiments across a synthetic basis function decomposition, multi-agent Craftax, and a continuous autonomous driving simulator (Highway-Env) confirm that GRID successfully disentangles reward structure in a semantically meaningful way, outperforms standard learning from demonstration baselines, and enables more efficient and stable specialization.

05.
arXiv (CS.AI) 2026-06-16

Mask-Proof: An LLM-based Automated Data Curation Pipeline on Mathematical Proofs

arXiv:2606.15258v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly capable of mathematical problem solving and can even assist with research-level proofs, yet we still lack a scalable and reproducible way to measure step-level reasoning in long proofs across diverse sources. This evaluation gap limits trustworthy AI assistance in proof-certified scientific progress. Existing evaluations often emphasize final answers or rely on costly expert grading, while end-to-end proof generation remains open-ended and hard to verify automatically. We introduce Mask-Proof, a pipeline that turns real proofs into automatically checkable masked-step tasks. It masks key formula steps, provides the necessary surrounding context, and evaluates model reconstructions with an LLM-based equivalence judge using repeated votes for stability. The resulting Mask-ProofBench contains 292 curated problems across diverse research areas. Experiments with 17 models show that reasoning-enhanced models outperform standard models by 12% to 27%. Our evaluator achieves 96.8% agreement with expert annotators, enabling faithful, reproducible, and comparable measurement of step-level mathematical reasoning. Benchmark, annotations, and code are available at https://github.com/weating/Mask-Proof.

06.
bioRxiv (Bioinfo) 2026-06-16

Better data, better trees: GenBank-GISAID deduplication and source-specific artifact masking in viral genomics

GenBank and GISAID are the primary repositories for viral genomic data, but integrating records across them remains a challenge. The same sequence could be made available in both databases without any cross-reference linking the two entries. Consequently, there is no systematic way to identify this redundancy, which compromises the compilation of representative, non-redundant large-scale datasets. In parallel, the growth of viral genomic data has increased the risk of systematic technical artifacts introduced during sequencing or assembly. These artifacts can inflate substitution rate estimates and degrade temporal signal, biasing evolutionary rate estimates. To address both challenges, here we present a formal, reproducible workflow integrating two newly developed complementary tools: G2G matcher for cross-repository harmonization and Lab-Specific Bias FILTer (LSBFILT) for masking of laboratory-specific artifacts. Using the Eastern/Central/South African (ECSA) chikungunya virus lineage as a proof-of-concept, we demonstrate that our integrated workflow restores temporal signal and provides a robust, curated dataset for downstream phylodynamic analyses. Critically, restricting masking of homoplastic sites to specific sequences reduces the substitution rate estimate from an inflated 8.517 x 10e-4; to 5.078 x 10e-4; substitutions/site/year and increases the coefficient of determination (R2) of the root-to-tip regression analysis from 0.353 to 0.677. By enabling systematic cross-repository harmonization and source-specific artifact masking, we provide the molecular epidemiological community with scalable tools to reconcile fragmented genomic data and reduce technical biases, fostering more accurate and reproducible phylogenetic analysis. G2G matcher is available at https://github.com/andrezaleite/G2G-Matcher, and LSBFILT at https://github.com/khourious/LSBFILT.

07.
medRxiv (Medicine) 2026-06-11

Advancing Clinical Implementation of Cardiovascular Polygenic Risk Scores Through Patient-Level Robustness Assessment

Background and Aims: Polygenic risk scores (PRSs) for atherosclerotic cardiovascular disease (ASCVD) can perform equivalently at the population level yet disagree for individual patients. We examined whether such intra-individual variability reflects genuinely complementary risk information or mainly statistical and methodological uncertainty, and whether it affects clinical classification once PRSs are integrated into SCORE2-OP. Methods: In 4,137 ASCVD-free participants of the CoLaus|PsyCoLaus cohort (478 incident events over a median 14.4 years), we identified 16 ASCVD-PRSs with practically equivalent population-level performance using Bayesian equivalence testing. We quantified intra-individual variability (standard deviation, coefficient of variation, intraclass correlation, Cohen's kappa, extreme discordance), tested whether discordance exceeded chance, decomposed scores into shared and unique genetic components, and assessed variability after integration into SCORE2-OP, benchmarked against perturbation of systolic blood pressure. Results: For a typical individual, risk estimates varied by 18 percentile points across PRSs. Discordance matched chance expectations under a shared-signal model, with no distinct phenotypic profile among discordant individuals, and predictive power resided overwhelmingly in the shared genetic component. Variability tracked PRS size and weighting rather than distinct variants. After integration into SCORE2-OP, 75.6% of participants were placed in different categories by at least one model and 54.6% as both low and high risk; instability was concentrated near guideline thresholds and far exceeded that from blood-pressure measurement error. Conclusions: Equivalent population-level performance is not sufficient to treat PRSs as interchangeable at the individual level, and methodological standardisation and pragmatic clinical trials remain necessary to determine whether PRS integration improves long-term cardiovascular outcomes.

08.
arXiv (CS.AI) 2026-06-15

The Perceived Fragility of Explanations in Audio Models: Manipulation of Attribution with Unchanged Predictions

arXiv:2606.14466v1 Announce Type: cross Abstract: This paper investigates the fragility of post-hoc explanation methods in audio deepfake detection. While previous work on explanation manipulation focused on images using standard $L_p$ metrics, we introduce a psychoacoustic framework that optimizes inaudible perturbations to decouple model attributions from final classifications. We evaluate this vulnerability across state-of-the-art architectures under strict prediction-preserving constraints. By evaluating the manipulation cost through domain-specific perceptual audio quality metrics alongside explanation alignment criteria, our framework demonstrates that an adversary can systematically distort automated explanation heatmaps while preserving the predicted deepfake label. Full code available at: https://github.com/cncPomper/Audio-XAI

09.
arXiv (CS.AI) 2026-06-12

MARS: Margin-Adversarial Risk-controlled Stopping for Parallel LLM Test-time Scaling

arXiv:2606.12935v1 Announce Type: new Abstract: Parallel test-time scaling samples many reasoning traces and majority-votes their answers, improving LLM accuracy but requiring traces to run to completion, incurring substantial computational overhead. We observe that probing partial traces at intermediate checkpoints can extract current answers without disrupting generation, revealing an evolving aggregate vote. Based on this observation, we introduce MARS, a margin-adversarial stopping rule that estimates which active traces are likely to change their answers and stops once the leader remains safe under a conservative bound on future vote movement. The rule separates two sources of uncertainty. It learns the trace-level switch probabilities that determine how much of the current margin is likely to be retained, while handling the harder question of where switching traces land through an adversarial bound calibrated from warmup traces. With true switch probabilities, MARS guarantees with high probability that the early-stopped answer matches the full-budget vote. In practice, a five-feature logistic model closely matches oracle switching behavior. Across three reasoning models and three competition-math benchmarks, MARS saves 25-47% of self-consistency tokens and 14-29% on top of DeepConf Online, a strong confidence-weighted baseline that already filters and truncates weak traces, while matching the accuracy of the corresponding full-budget baselines.

10.
arXiv (CS.AI) 2026-06-16

Multi-Granular Node Pruning for Causal Circuit Discovery

arXiv:2512.10903v2 Announce Type: replace Abstract: Circuit discovery aims to identify minimal subnetworks that are responsible for specific behaviors in large language models (LLMs). Existing approaches primarily rely on iterative edge pruning, which is computationally expensive and limited to coarse-grained units such as attention heads or MLP blocks, overlooking finer structures like individual neurons. We propose a node-level pruning framework for circuit discovery that addresses both scalability and granularity limitations. Our method introduces learnable masks across multiple levels of granularity, from entire blocks to individual neurons, within a unified optimization objective. Granularity-specific sparsity penalties guide the pruning process, allowing a comprehensive compression in a single fine-tuning run. Empirically, our approach identifies circuits that are smaller in nodes than those discovered by prior methods; moreover, we demonstrate that many neurons deemed important by coarse methods are actually irrelevant, while still maintaining task performance. Furthermore, our method has a significantly lower memory footprint, 5-10x, as it does not require keeping intermediate activations in the memory to work.

11.
arXiv (CS.CL) 2026-06-16

Your "Pro" LLM Subscription May Actually Be "Free": Exposing Fingerprint Spoofing Risks in LLM Inference Services

As Large Language Model (LLM) APIs become ubiquitous, users increasingly rely on black-box fingerprinting to verify that providers are serving the advertised premium models. However, these methods may overlook adversarial providers who manipulate model weights to cheat the fingerprint process. We introduce a novel threat termed fingerprint spoofing, where a malicious provider stealthily serves a weaker model that has been parameter-efficiently fine-tuned to mimic a stronger model, thereby evading user-side fingerprinting. We first formally prove that user-side resource constraints (i.e., finite query budgets and weak fingerprinting classifiers) make current fingerprinting vulnerable to fingerprint spoofing. Guided by this theoretical analysis, we propose GhostPrint, a cost-effective attack framework leveraging surrogate modeling, reward-ranked fine-tuning, and knowledge distillation. Extensive evaluations in both static and continual fingerprinting settings demonstrate that GhostPrint allows weak models to consistently bypass representative fingerprint methods while maintaining utility at a low fine-tuning cost, exposing a critical vulnerability in current LLM fingerprinting pipelines.

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

Diffusion Integrated Gradients: Controllable Path Generation for Flexible Feature Attribution

arXiv:2606.22314v2 Announce Type: replace-cross Abstract: Path-based attribution methods such as Integrated Gradients (IG) are widely adopted for their strong axiomatic properties and effectiveness in attributing model predictions to input features by integrating gradients along a path from a baseline to the input. However, the choice of the attribution path largely affects the quality of explanations, and existing approaches rely on fixed or hand-crafted paths that often produce noisy or distorted attributions. To address this limitation, we propose Diffusion Integrated Gradients (DiffIG), a novel method that reformulates path generation as a conditional generative modeling problem. DiffIG first trains a diffusion model to learn a distribution over paths generated from a Stick-Breaking Process, then employs guided sampling to embed user guidance during the sampling procedure. We demonstrate that DiffIG quantitatively matches or outperforms existing path-based methods, achieving perceptually aligned explanations. This work introduces a new generative perspective for flexible, inference-time controllable Explainable Artificial Intelligence (XAI) methods.

13.
arXiv (CS.CV) 2026-06-19

TriFlow: Generating Artist-Like 3D Mesh Topology via Nearest-Vertex Vector Fields

We present TriFlow, a new generative approach for producing compact 3D meshes with artist-like triangle topology directly from input geometry conditions such as signed distance fields. Our key insight is to represent mesh topology as a nearest-vertex vector field (NVF) defined over the surface, where each point encodes its association to the nearest triangle vertex in the local barycentric frame. We train a latent flow-matching model to synthesize this field, enabling topology generation conditioned on the input geometry. To extract a coherent mesh, we cluster surface regions using the generated NVF and guide a constrained quadric error metric (QEM) mesh simplification with topology-aware optimization. This yields output meshes that closely match the input geometry while exhibiting structured, artist-like connectivity. Experiments demonstrate that TriFlow achieves stronger generalization and significantly improved topology quality compared to state-of-the-art learning-based approaches, alongside 90% lower Chamfer Distance and an 8x speedup.

14.
bioRxiv (Bioinfo) 2026-06-19

HTS-Oracle v2: Prospective AI-Guided Discovery and Experimental Validation of Small Molecule Modulators Across Multiple Targets

High-throughput screening (HTS) remains the cornerstone of early-phase small molecule discovery yet consistently underperforms against immunotherapy targets, yielding validated hit rates below 0.1%. Here we introduce HTS-Oracle v2, which features rigorous cross-validation that ensures honest performance estimates. HTS-Oracle v2 was trained and validated across four clinically significant immune checkpoint targets (CD28, ICOS, LAG-3, and TIGIT) achieving ROC-AUC values of 0.968, 0.969, 0.875, 0.928 respectively under rigorous cross-validation. For prospective experimental validation, HTS-Oracle v2 was applied to an 8,960-compound Enamine Protein Mimetic Library, selecting only 25 compounds per target for experimental testing using temperature-related intensity change (TRIC) technology, a 99.7% reduction in screening burden. HTS-Oracle v2 identified 4, 5, 4, and 6 validated binders from 25 prospectively selected compounds per target, corresponding to validated hit rates of 16%, 20%, 16%, and 24%, respectively. Notably, 67-80% of all experimentally confirmed hits across the full 8,960-compound library were captured within just 25 model-selected compounds per target. For CD28, this represents a 28-fold improvement over HTS-Oracle v1 (239x versus 8.4x), establishing HTS-Oracle v2 as an efficient platform for AI-guided prospective hit discovery across immunotherapy targets.

15.
arXiv (CS.CV) 2026-06-16

Seeing Roads Through Words: A Language-Guided Framework for RGB-T Driving Scene Segmentation

Robust semantic segmentation of road scenes under adverse illumination, lighting, and shadow conditions remain a core challenge for autonomous driving applications. RGB-Thermal fusion is a standard approach, yet existing methods apply static fusion strategies uniformly across all conditions, allowing modality-specific noise to propagate throughout the network. Hence, we propose CLARITY that dynamically adapts its fusion strategy to the detected scene condition. Guided by vision-language model (VLM) priors, the network learns to modulate each modality's contribution based on the illumination state while leveraging object embeddings for segmentation, rather than applying a fixed fusion policy. We further introduce two mechanisms - one which preserves valid dark-object semantics that prior noise-suppression methods incorrectly discard, and a hierarchical decoder that enforces structural consistency across scales to sharpen boundaries on thin objects. Experiments on the MFNet dataset demonstrate that CLARITY establishes a new state-of-the-art (SOTA), achieving 62.3% mIoU and 77.5% mAcc.

16.
medRxiv (Medicine) 2026-06-10

Trajectories of brain structure and function in young adult carriers of genetic frontotemporal dementia variants

Background and Objectives: Converging evidence hints at neurodevelopmental effects in genetic frontotemporal degeneration (FTD). In cross-sectional studies, for some genes, young adult FTD variant carriers show differences in brain volumes and cognition compared to familial non-carriers. However, longitudinal trajectories may more sensitively capture FTD-related neurodevelopmental vs. neurodegenerative changes than cross-sectional approaches. This study examined longitudinal trajectories of brain volumes, executive function, and plasma biomarkers in young adult carriers compared to familial non-carriers, as measures of neurodevelopmental and neurodegenerative outcomes of FTD-causing variants. Methods: This longitudinal cohort study comprised participants, aged 18-30 years, from the FTD Prevention Initiative across Europe, Canada, and the USA. Genetic groups included C9orf72 (47%), MAPT (30%), and GRN (23%). Linear mixed-effects models were computed to assess longitudinal outcomes across age between groups, controlling for sex, scanner (for brain volumes), and education (for executive function); random effects accounted for between-subject variability nested within family membership. Results: Variant carriers (n=147) and familial non-carriers (n=113) did not differ in age (mean{+/-}SD, 25.9{+/-}3.2 years), sex (53% female), or number of visits (2.1{+/-}1.7). Young adult C9orf72 repeat expansion carriers exhibited smaller thalamic volumes than non-carriers at the reference age of 26 years (b=-982.8mm3, SE=317.0, p=0.0046, f2=0.32), with relatively stable trajectories across ages 18-30 (i.e., no change over time). Trajectories of rostral anterior cingulate volumes differed in C9orf72 carriers and non-carriers across age, where carriers showed relatively stable trajectories and non-carriers showed age-appropriate declines (b=64.4mm3, SE=29.9, p=0.035, f2=0.07). For MAPT and GRN, there were little to no differences in total brain, cortical, or subcortical volumes between groups and over time. No longitudinal differences were observed between carriers and non-carriers in executive function, or plasma NfL or GFAP for any genetic group. Discussion: C9orf72 repeat expansions were linked to smaller average thalamic volumes and stable trajectories between ages 18 to 30, supporting potential neurodevelopmental origins. The modest evidence supporting an absence of difference in neurodegenerative biomarkers and executive function suggests minimal early neurodegeneration and functional preservation in young adulthood.

17.
arXiv (CS.CL) 2026-06-24

Quantifying Prior Dominance in RAG Systems

作者:

Retrieval-Augmented Generation (RAG) grounds Large Language Models in external knowledge, yet current evaluations rely on discrete heuristics that suffer from ''epistemic blindness'' - failing to distinguish genuine contextual information extraction from parametric memory recall. To address this, we introduce the Normalized Context Utilization (NCU) metric, leveraging continuous token log-probabilities across zero-shot, oracle, and adversarial conditions to strictly quantify contextual information gain. Evaluating architectures ranging from 1.5B to 72B parameters alongside a proprietary commercial API reveals that for strict factual extraction (without Chain-of-Thought reasoning), traditional scaling laws exhibit extreme diminishing returns: highly efficient Small Language Models (SLMs) match or outperform high-capacity architectures. Furthermore, we demonstrate that ``Prior Dominance'' correlates with model scale and proprietary alignments. The evaluated commercial API not only overrode explicit external evidence in nearly half of adversarial conflicts, but also frequently suffered from systemic confidence collapse (Negative Transfer) when its parametric priors were contradicted. Our findings highlight the structural epistemic advantage and superior contextual adherence of SLMs in strict extraction workflows.

18.
arXiv (CS.CV) 2026-06-15

3D-RFT: Reinforcement Fine-Tuning for Video-based 3D Scene Understanding

Reinforcement Learning with Verifiable Rewards ( RLVR ) has emerged as a transformative paradigm for enhancing the reasoning capabilities of Large Language Models ( LLMs), yet its potential in 3D scene understanding remains under-explored. Existing approaches largely rely on Supervised Fine-Tuning ( SFT), where the token-level cross-entropy loss acts as an indirect proxy for optimization, leading to a misalignment between training objectives and task performances. To bridge this gap, we present Reinforcement Fine-Tuning for Video-based 3D Scene Understanding (3D-RFT ), the first framework to extend RLVR to video-based 3D perception and reasoning. 3D-RFT shifts the paradigm by directly optimizing the model towards evaluation metrics. 3D-RFT first activates 3D-aware Multi-modal Large Language Models ( MLLM s) via SFT, followed by reinforcement fine-tuning using Group Relative Policy Optimization ( GRPO) with strictly verifiable reward functions. We design task-specific reward functions directly from metrics like 3D IoU and F1-Score to provide more effective signals to guide model training. Extensive experiments demonstrate that 3D-RFT-4B achieves state-of-the-art performance on various video-based 3D scene understanding tasks. Notably, 3D-RFT-4B significantly outperforms larger models (e.g., VG LLM-8B) on 3D video detection, 3D visual grounding, and spatial reasoning benchmarks. We further reveal good properties of 3D-RFT such as robust efficacy, and valuable insights into training strategies and data impact. We hope 3D-RFT can serve as a robust and promising paradigm for future development of 3D scene understanding.

19.
arXiv (CS.LG) 2026-06-24

Grouped Query Experts: Mixture-of-Experts on GQA Self-Attention

arXiv:2606.20945v2 Announce Type: replace Abstract: Self-attention is central to Transformer performance and is often the most expensive part of the Transformer at long context lengths because its pairwise token interactions scale quadratically with sequence length. Standard dense attention also applies the same set of attention heads to every token regardless of token difficulty or information content. This uniform activation can waste compute, especially as sequences grow longer and attention cost increases rapidly. We propose Grouped Query Experts (GQE), a mixture-of-experts layer on top of grouped-query attention (GQA). Within each GQA group, a router selects k query-head experts per token while all key-value (KV) heads remain dense and unchanged. Thus, GQE keeps the KV cache benefits of GQA and reduces only the active query-head computation. On a fixed 30B token budget at the 250M parameter scale, GQE matches the all-active GQA baseline in downstream accuracy while activating half the query heads per token.

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

Toward Preference-aligned Large Language Models via Residual-based Model Steering

Preference alignment is a critical step in making Large Language Models (LLMs) useful and aligned with (human) preferences. Existing approaches such as Reinforcement Learning from Human Feedback or Direct Preference Optimization typically require curated data and expensive optimization over billions of parameters, and eventually lead to persistent task-specific models. In this work, we introduce Preference alignment of Large Language Models via Residual Steering (PaLRS), a training-free method that exploits preference signals encoded in the residual streams of LLMs. From as few as one hundred preference pairs, PaLRS extracts lightweight, plug-and-play steering vectors that can be applied at inference time to push models toward preferred behaviors. We evaluate PaLRS on various small-to-medium-scale open-source LLMs, showing that PaLRS-aligned models achieve consistent gains on mathematical reasoning and code generation benchmarks while preserving baseline general-purpose performance. Moreover, when compared to models aligned with DPO and SimPO, they perform better with great time-savings. Our findings highlight that PaLRS offers an effective, much more efficient and flexible alternative to standard preference optimization pipelines, offering a training-free, plug-and-play mechanism for alignment with minimal data.

21.
arXiv (CS.LG) 2026-06-24

Bridging Mechanistic Interpretability and Prompt Engineering with Gradient Ascent for Interpretable Persona Control

arXiv:2601.02896v3 Announce Type: replace Abstract: Controlling emergent behavioral personas (e.g., sycophancy, hallucination) in Large Language Models (LLMs) is critical for AI safety, yet remains a persistent challenge. Existing solutions face a dilemma: manual prompt engineering is intuitive but unscalable and imprecise, while automatic optimization methods are effective but operate as "black boxes" with no interpretable connection to model internals. We propose a novel framework that adapts gradient ascent to LLMs, enabling targeted prompt discovery. In specific, we propose two methods, RESGA and SAEGA, that both optimize randomly initialized prompts to achieve better aligned representation with an identified persona direction. We introduce fluent gradient ascent to control the fluency of discovered persona steering prompts. We demonstrate RESGA and SAEGA's effectiveness across Llama 3.1, Qwen 2.5, and Gemma 3 for steering three different personas, sycophancy, hallucination, and myopic reward. Crucially, on sycophancy, our automatically discovered prompts achieve significant improvement (49.90% compared with 79.24%). By grounding prompt discovery in mechanistically meaningful features, our method offers a new paradigm for controllable and interpretable behavior modification. We release our scripts for RESGA and SAEGA in this github repo: https://github.com/HarshSaini10/RESGA_SAEGA.

22.
arXiv (CS.AI) 2026-06-19

Formal Verification of Learned Multi-Agent Communication Policies via Decision Tree Distillation

arXiv:2606.19632v1 Announce Type: cross Abstract: Multi-agent reinforcement learning (MARL) enables agents to develop coordination strategies through emergent communication, but neural policies lack the formal safety guarantees required for safety-critical robotic deployment in drone swarms and autonomous vehicle fleets. We present the first end-to-end framework for safety verification of learned multi-agent communication policies through policy abstraction: neural policies are distilled into interpretable decision trees, then formally verified, with empirical validation confirming that verified safety properties transfer to original networks. Our four-stage pipeline consists of domain-specific feature extraction from agent observations, decision tree distillation achieving 97.9% +/- 1.2% fidelity to neural policies, automated translation to PRISM probabilistic model checker specifications with complete feature-to-state-variable correspondence, and compositional verification of Probabilistic Computation Tree Logic (PCTL) properties via pairwise decomposition with union-bound aggregation and empirical neighbor modeling. Evaluating Vector-Quantized Variational Information Bottleneck (VQ-VIB) policies for multi-drone coordination with 5-7 agents, we verify 18 temporal logic properties across safety, liveness, and cooperation, achieving 88.9% property satisfaction with all five safety thresholds satisfied (0.3% collision probability vs. 1% threshold). Monte Carlo validation of original neural policies confirms that verified safety properties transfer with

23.
medRxiv (Medicine) 2026-06-15

Repurposing cardiovascular disease risk models to predict incident and co-occurring cardiovascular, cardiometabolic and neurocognitive outcomes.

Background: Cardiovascular disease (CVD), cardiometabolic and neurocognitive conditions share risk factors and frequently co-occur. We evaluated whether four established CVD risk prediction models (QRISK3, PCE, SCORE2, SCORE2-OP) can be repurposed to predict 10-year risk of these conditions and their co-occurrence with CVD. Methods: The models were recalibrated using 20% of the UK Biobank (UKB) and evaluated in the remaining 80%. We performed external validation using data from Clinical Practice Research Datalink (CPRD) Aurum, assessing model discrimination (c-statistics) and calibration (intercept and slope). We used permuted feature importance to determine the influence of each individual predictor in the models. Results: Depending on the model, the c-statistics for incident CVD ranged from 0.71 to 0.74 in the UKB test set (16,137 events). Discrimination was equal to or higher than CVD when evaluated against non-traditional CVD outcomes: 0.74 to 0.77 for heart failure (3,471 events), 0.72 to 0.73 for atrial fibrillation (9,213 events), 0.73 to 0.75 for peripheral arterial disease (1,927 events) and 0.80 to 0.82 for abdominal aortic aneurysm (595 events). For the multimorbidity endpoints, model discrimination ranged from 0.74 for the composite of CVD and T2DM (SCORE2-OP) to 0.83 for the composite of CVD and dementia or Parkinson's disease (QRISK3). When considering the onset of any cardiovascular, cardiometabolic, or neurocognitive outcome discrimination ranged from 0.71 to 0.72. The repurposed models slightly underestimated the predicted risk in the CPRD compared to the UKB: average difference in calibration intercept was at most -0.64. After age and sex, smoking status and systolic blood pressure contributed most to model predictions. Conclusions: Repurposed CVD models can be used to identify 10-year risk of many CVD-related conditions and their multimorbidity. These may be used to support risk-based approaches to prevention and screening. The repurposed models have been made available at: https://repurposed-cvd-risk-models.shinyapps.io/cvd_cmd_dementia_app/ Keywords: Risk prediction; cardiovascular disease; cardiometabolic disease; dementia; disease prevention.

24.
medRxiv (Medicine) 2026-06-12

Disentangling Confounders from Pathology in Long-COVID Trajectory Prediction for Women: An Interpretable Large-Language-Model Approach

Objective. Post-acute sequelae of SARS-CoV-2 infection (PASC, "Long COVID") dispropor- tionately affects women, in whom hallmark symptoms–insomnia, fatigue, palpitations, cogni- tive difficulty–overlap with comorbidities and hormonal transitions such as menopause. This diagnostic overlap is a confounding problem: models that forecast future symptom severity risk attributing baseline physiological noise to viral pathology. We ask whether an interpretable, causally disentangled language model can separate true pathological signal from such con- founders while remaining competitive with strong predictors of future PASC severity

25.
arXiv (CS.AI) 2026-06-16

ARB4WM: An Adversarial Robustness Benchmark for World Models in Continuous Control

arXiv:2606.16605v1 Announce Type: new Abstract: World models are widely used in robotic and agentic engineering control systems due to their ability to learn latent dynamics for planning and decision-making. As these systems are increasingly deployed in safety-critical settings, understanding their robustness under adversarial conditions has become essential. However, existing evaluations lack a unified benchmark for testing adversarial threats across the policy, value, and latent-dynamics levels of world-model agents. To fill this gap, we present ARB4WM, a unified evaluation framework for pre-deployment robustness and risk assessment of world-model agents under visual perturbations. ARB4WM defines five white-box loss objectives across these three levels and studies their effects when combined with single-step or multi-step perturbation strategies and temporal attack modes, including full-frame, half-sequence, and sparse-frame exposure. Specifically, we evaluate four Dreamer-style agents across 20 tasks from MetaWorld and the DeepMind Control Suite under different loss objectives, perturbation strategies, and temporal attack modes. Results show that attacks targeting value estimation, latent representations, and RSSM dynamics can be as damaging as direct policy disruption, and that early or frequent perturbations are especially harmful, while input-level defenses provide limited recovery under adaptive attacks. These findings suggest that safety, risk, and reliability assessment for world models should cover multiple component-oriented attack objectives and temporal exposure protocols rather than relying solely on action-space robustness. Source code is available at https://github.com/zaoanguai/ARB4WM.