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.CL) 2026-06-15

Deep Dense Exploration for LLM Reinforcement Learning via Pivot-Driven Resampling

Effective exploration is a key challenge in reinforcement learning for large language models: discovering high-quality trajectories within a limited sampling budget from the vast natural language sequence space. Existing methods face notable limitations: GRPO samples exclusively from the root, saturating high-probability trajectories while leaving deep, error-prone states under-explored. Tree-based methods blindly disperse budgets across trivial or unrecoverable states, causing sampling dilution that fails to uncover rare correct suffixes and destabilizes local baselines. To address this, we propose Deep Dense Exploration (DDE), a strategy that focuses exploration on $pivots$-deep, recoverable states within unsuccessful trajectories. We instantiate DDE with DEEP-GRPO, which introduces three key innovations: (1) a lightweight data-driven utility function that automatically balances recoverability and depth bias to identify pivot states; (2) local dense resampling at each pivot to increase the probability of discovering correct subsequent trajectories; and (3) a dual-stream optimization objective that decouples global policy learning from local corrective updates. Experiments on mathematical reasoning benchmarks demonstrate that our method consistently outperforms GRPO, tree-based methods, and other strong baselines. Code is available at https://github.com/AgentCombo/DEEP-GRPO

02.
arXiv (quant-ph) 2026-06-12

Fibonacci Steady-States and Persistent Oscillations in an Ordered Multimode Dicke Model

arXiv:2606.13072v1 Announce Type: new Abstract: Ultracold atoms in multimode optical cavities provide a rich testbed for many-body phenomena enabled by light-mediated interactions. Recent experiments include realizations of spin glasses and associative memories, as described by multimode Dicke models with disordered couplings. However, the properties of multimode Dicke models with ordered coupling geometries remain largely unexplored. In this work, we investigate the stable steady-states of the multimode Dicke model with an ordered nearest-neighbor coupling geometry, where $n_c$ atomic clusters are coupled via $n_c-1$ cavity modes. We show that the number of mean-field stable steady-states in the superradiant phase exhibits Fibonacci scaling with the number of atomic clusters, and that a subset of these steady-states exhibit persistent oscillations. Using both the truncated Wigner approximation and the numerically-exact hierarchy of pure states, we further demonstrate that these features of the stable steady-state solutions persist for finite cluster sizes. Ordered multimode Dicke models, such as the nearest-neighbor coupling geometry considered here, are accessible with current experimental technologies and point toward a broader class of strongly interacting dissipative systems with similarly rich behavior.

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

GRAPE: Guided Parameter-Space Evolution for Compact Adversarial Robustness

arXiv:2606.14865v1 Announce Type: cross Abstract: Adversarial Training (AT) improves neural network robustness, but most methods train a fixed parameter space from the start. This paper asks whether the order in which parameters become optimizable can affect the final robust solution, even when the final architecture or computation budget is controlled. We propose GRAPE, Guided Parameter-Space Evolution, a training framework for compact adversarial robustness. GRAPE combines parameter-space stabilization with progressive hidden expansion: it stabilizes robust optimization in the currently exposed space, gradually releases new optimizable dimensions, and uses an adversarial spectral utilization score to guide newly released capacity toward high-pressure modules. In contrast to fixed-structure AT, GRAPE treats robust model learning as a process of progressive parameter-space exposure and evolution. Under the standard $\ell_\infty$ threat model on CIFAR-10, with fixed-structure ResNet-18 AT as a controlled reference, GRAPE improves PGD-20 robust accuracy from 51.70% to 56.94% at a nearly matched computation budget with a FLOPs ratio of 1.009x, while reducing parameter count by about 21.4%. A sequential grow variant with the same final ResNet-18 architecture reaches 56.52% PGD-20 robust accuracy, indicating that the gain is not only due to final architecture differences but also to the parameter-space exposure path. These results suggest that guided parameter-space evolution can yield compact and robust parameter configurations under matched computation.

04.
arXiv (CS.CV) 2026-06-24

Training-Time Optical Priors for Wireless Capsule Endoscopy Classification: Hemoglobin-Aware Input Fusion with Cross-Vendor Evaluation

Background. RGB-trained classifiers for wireless capsule endoscopy (WCE) conflate hemoglobin contrast with bile staining and illumination falloff, limiting sensitivity to small-vessel vascular findings such as Lymphangiectasia. We introduce a physics-informed framework that injects an analytic, Monte-Carlo-inspired hemoglobin prior into a standard classifier purely at training time – to our knowledge the first use of an explicit optical light-transport prior in WCE classification. Methods. On Kvasir-Capsule (47,238 frames, 43 patients, 11 evaluable classes; patient-disjoint split) we test, across 6 seeds against an RGB-only EfficientNet-B0 baseline: (i) a 5-channel input-fusion variant feeding the prior P_blood alongside RGB; (ii) a distillation variant that runs on plain 3-channel RGB at inference; and (iii) a three-stream extension adding a temporal Transformer and an autoencoder-residual stream. We replicate across ResNet-18 and ConvNeXt-Tiny and report cross-vendor zero-shot transfer on the public Galar cohort. Results. Input fusion lifts cross-seed macro-AUC 0.760 -> 0.783 (5/6 seeds positive); distillation reaches 0.773; the three-stream model reaches 0.804 (+0.044 over baseline, paired DeLong p < 1e-4). Lymphangiectasia AUC rises 0.238 -> 0.337, sign-consistent across all 6 seeds. A four-variant ablation reveals a parameterization-mechanism boundary: only the spatial-channel form lifts. Cross-vendor zero-shot on Galar retains ~60% of the ConvNeXt-Tiny lift.

05.
arXiv (CS.CV) 2026-06-24

MILE: A Mechanically Isomorphic Hand Exoskeleton and Visuotactile Robotic Hand for Data Collection in Dexterous Manipulation

Dexterous robotic hands are expected to perform complex, contact-rich object manipulation, but learning such skills remains challenging because high-dimensional hands require high-fidelity demonstrations. Imitation learning provides a practical route for acquiring dexterous manipulation skills from human demonstrations, yet collecting synchronized multimodal demonstrations with accurate hand actions and tactile observations remains a key bottleneck. We present MILE, a teleoperation-based data-collection system comprising the human-first MILE exoskeleton and the mechanically corresponding MILE-Tac robotic hand. The system integrates custom-designed and fabricated modular joint encoders and compact MILE fingertip visuotactile sensor modules. The exoskeleton is informed by human-hand anatomy and ergonomic constraints, while the robotic hand is co-designed to preserve the selected four-finger kinematic topology. This correspondence enables joint-space command transfer and reduces reliance on task-space IK-based retargeting. The system synchronously records task-specific visual observations, four fingertip visuotactile streams, robot-hand proprioception, and exoskeleton-derived action commands. We evaluate MILE through a four-task teleoperation benchmark against representative glove-based and vision-based interfaces, and through imitation-learning experiments that compare policies trained with and without fingertip tactile input. The project page is available at https://sites.google.com/view/mile-system.

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

If These Walls Could Talk: Critical Play with Large Language Models in Museums

arXiv:2606.15565v1 Announce Type: cross Abstract: Large Language Models (LLMs) are increasingly being used in museums to as role playing chatbots which let visitors talk to simulated versions of people and artefacts from the past. While such installations can be playful and engaging, they are also problematic because LLMs cannot be trusted to speak truthfully. I identify a fundamental dilemma for the use of LLMs in museum chatbots: LLMs cannot be trusted to tell the truth, and efforts to make them more reliable may ruin that which is attractive about the bots in the first place - their ability to engage in life-like conversation. In response, I propose designing for critical play with LLM-based bots: Designing for playful interactions with bots that are unreliable but still able to represent the past in an adequate and engaging manner - as fictional characters representing historical narratives, styles of discourse, diverse perspectives, humor and satire.

07.
arXiv (CS.CV) 2026-06-24

S1-Omni-Image: A Unified Model for Scientific Image Understanding, Generation, and Editing

We present S1-Omni-Image, an open-weight unified multimodal model for scientific image understanding, generation, and editing. Unlike general-purpose image generation models, scientific image tasks require not only high-fidelity synthesis, but also robust understanding of scientific semantics, structural relations, domain knowledge, and task intent. To this end, S1-Omni-Image builds on the scientific multimodal reasoning backbone S1-VL-32B and couples its understanding capability with an image generation module under a unified think-before-generate paradigm. Given a user instruction, the model first produces a task-oriented reasoning trace, a textual answer, and a task special token; their hidden states are then injected into the generation module to condition image generation or editing. S1-Omni-Image supports scientific image understanding, generation, and editing in a unified framework. For generation, it focuses on scientific illustrations and text rendering, including logical diagrams, relational comparisons, data charts, and realistic scientific visualizations. For editing, it casts segmentation and other domain-specific vision tasks as native image editing problems, enabling multi-turn illustration editing, medical and geographic image segmentation, medical image translation, and scientific image super-resolution. We construct SciGenEdit, a 314K-sample training dataset, and release the model weights, inference code, and SciGenEdit-10K. Experiments show that S1-Omni-Image substantially improves scientific image generation and editing while preserving the scientific image understanding capability inherited from S1-VL-32B. It outperforms open-source models on GenExam and TechImage-Bench, achieves state-of-the-art results on four editing benchmarks including MSD, cigRockSEM, SynthRAD2025, and IXI, and maintains stable performance on scientific image understanding evaluations.

08.
arXiv (CS.LG) 2026-06-18

ThousandWorlds: A benchmark for climate emulation of potentially habitable exoplanets

arXiv:2606.18338v1 Announce Type: new Abstract: The search for life beyond Earth will depend on detecting faint signatures in the atmospheres of potentially habitable exoplanets. Interpreting those signatures requires understanding the host planet's climate: the same molecule may signal life on one planet and abiotic chemistry on another. Global climate models (GCMs) provide this understanding, but individual runs can require up to millions of core-hours and substantial domain expert time. Machine-learning emulators could remove this bottleneck, but progress has been limited by the absence of a curated, multi-model exoclimate dataset. We introduce ThousandWorlds, an ML-ready benchmark for exoclimate emulation and for the broader regime of low-data, multi-simulator, parameter-to-field regression. The dataset contains approximately 1800 simulations from five GCMs, mapping eight planet parameters to 3D atmospheric fields including temperature, humidity, winds, clouds, and radiation. Three nested subsets define progressively harder challenges: single-simulator regression, multi-simulator regression with complete observations, and multi-simulator regression with structured missingness. We propose two evaluation protocols: one for ranking methods, and one that measures performance relative to the disagreement between GCMs themselves. We evaluate seven baselines spanning simple methods, deep learning, and Gaussian processes. GP-based methods perform best, suggesting that ThousandWorlds exposes a regime where off-the-shelf deep learning does not yet succeed. Data: https://doi.org/10.57967/hf/8695. Code: https://github.com/edstevenson/ThousandWorlds.

09.
arXiv (CS.CL) 2026-06-18

Breaking the Solver Bottleneck: Training Task Generators at the Learnable Frontier

The limiting resource for training agents via reinforcement learning (RL) is increasingly frontier task supply: valid, solvable tasks just difficult enough to train the current model. As reasoning and agentic models improve, fixed task distributions saturate, while naive synthetic generation yields tasks that are trivial, impossible, or ill-posed. Training a task generator with RL to optimize validity and learnability can address this bottleneck, but direct optimization requires repeated solver rollouts per candidate. For software-engineering (SWE) tasks, a single rollout can take tens of minutes; solver-in-the-loop generator training is intractable. We introduce PROPEL, a solver-amortized framework for training task generators at the targeted solve rate. PROPEL trains a lightweight activation probe on a one-time labeled corpus of generated tasks and solver outcomes. The probe predicts target-solver pass rate from a frozen generator reference model and serves as a proxy for solve rate during generator optimization, reducing generator evaluation to a single forward pass. Across math, code, and software-engineering at multiple model scales, PROPEL shifts generation toward the targeted solve rate: for coding, tasks generated at the learnable frontier increase from $10.1\% \rightarrow 20.0\%$ for a Qwen2.5-3B-Instruct solver and from $5.3\% \rightarrow 12.6\%$ for a Qwen2.5-7B-Instruct solver. For SWE, PROPEL increases the share of generations at the targeted solve rate from $9.8\% \rightarrow 19.6\%$ for Qwen3.5-27B on repositories not seen during training of probe and generator.

10.
medRxiv (Medicine) 2026-06-15

CDH13 is associated with cellular viability after exposure to ionizing radiation using genome-wide screening

Background: It is well known that genetic variants contribute to cellular sensitivity to chemotherapeutic agents and ionizing radiation (IR). The aim of this study was to identify single nucleotide polymorphisms (SNPs) and genes associated with the spectrum of normal cellular sensitivity of lymphoblastoid cell lines (LCLs) towards ionizing radiation and mitomycin C (MMC). Methods: In a first step, we determined the viability of LCLs established from male participants of the Berlin Aging Study II (BASE-II) aged >=62 years following treatments with increasing doses of IR (n=137 cell lines) or MMC (n=140 cell lines) using the alamarBlue assay. Results from intra-experimental triplicates and three independent experiments for each cell line and treatment were used to calculate the area under the curves (AUCs) representing the specific sensitivity to IR and MMC of each LCL. The data from these experiments were subsequently used as outcomes in genome-wide association studies (GWASs). In addition, we calculated polygenic risk scores (PGS) from UK Biobank GWAS results for four cancer-related phenotypes and assessed the extent to which the variance in the IR and MMC sensitivity is explained by these PGS. Results: The GWAS analyses revealed one variant, rs74728080, located in CDH13 on chromosome 16, to show genome-wide significant (p < 5 x 10-8, beta = 2.81) association with cellular viability after treatment with IR. In the GWAS on MMC sensitivity the most interesting signal was elicited by SNP rs113978558 in an intron of the PLD5 gene on chromosome 1 (p = 9.232 x 10-8; beta = 1.44). Several other SNPs with statistically suggestive (i.e., p < 1 x 10-5) evidence of association with IR or MMC sensitivity were identified. PGSs calculations from GWAS of four cancer-related traits in UKB explained ~5% and ~3% of phenotypic variance in IR- and MMC-induced cell viability, respectively. Conclusion: The genome-wide significant association of rs74728080 with IR sensitivity and the location of this variant in CDH13 is interesting and functionally highly plausible given its known involvement in oxidative-stress response and function as tumor suppressor. Taken together, our novel data suggest that CDH13 may be genuinely involved in regulating cellular IR sensitivity.

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

Time-Varying Audio Effect Modeling by End-to-End Adversarial Training

arXiv:2512.15313v2 Announce Type: replace-cross Abstract: Deep learning has become a standard approach for the modeling of audio effects, yet strictly black-box modeling remains problematic for time-varying systems. Unlike time-invariant effects, training models on devices with internal modulation typically requires the recording or extraction of control signals to ensure the time-alignment required by standard loss functions. This paper introduces a Generative Adversarial Network (GAN) framework to model such effects using only input-output audio recordings, without requiring a modulation signal extraction. We propose a convolutional-recurrent architecture trained via a two-stage strategy: an initial adversarial phase allows the model to learn the distribution of the modulation behavior without strict phase constraints, followed by a supervised fine-tuning phase where a State Prediction Network (SPN) estimates the initial internal states required to synchronize the model with the target. Additionally, a new metric based on chirp-train signals is developed to quantify modulation accuracy. Experiments modeling a vintage hardware phaser demonstrate the method's ability to capture time-varying dynamics in a fully black-box context.

12.
arXiv (CS.LG) 2026-06-16

Peak-Based Nuclide Identification in HPGe $\gamma$-Spectrometry with Machine Learning and SHAP

arXiv:2606.14874v1 Announce Type: cross Abstract: High-purity germanium gamma spectra often require time-consuming analyses from subject matter experts. Photopeaks within these spectra are carefully fitted and numerical methods are employed to assist with nuclide identification (NID) and quantification. Amending the list of nuclides identified by analysis software can be nontrivial. When many samples need to be analyzed, it is therefore challenging to make timely and correct decisions. Supervised machine-learning-based NID can serve as an expert-informed, automated tool to improve the initial set of radionuclides suggested to an analyst and more effectively drive subsequent quantification. To that end, we implemented machine learning models that map photopeaks carefully fitted by analysts to NID results for experimental spectra containing various isotopic combinations drawn from a set of 65 isotopes. The best model achieved an F1 score of 0.97, markedly surpassing the F1 score of 0.84 achieved by traditional software when compared using a nuclide library comprising the same 65 isotopes assessed by the models. Finally, we illustrated the most important input features for model predictions using Shapley Additive Explanations. These explanations revealed that the models use physically relevant photopeaks when making predictions for the isotopes in our nuclide library.

13.
medRxiv (Medicine) 2026-06-17

Macrophage-targeted glucocorticoid prodrug resolves acute inflammation while preserving HPA axis function: mechanistic, preclinical, and Phase II/III clinical evidence

Glucocorticoids (GCs) remain the fastest-acting anti-inflammatory agents but are constrained by systemic exposure that suppresses the hypothalamic pituitary adrenal (HPA) axis, silences adaptive immunity, and drives chronic toxicities. Chronic inflammatory diseases are sustained by long-lived CD206+ macrophages containing immune-resistant pathogenic material not cleared physiologically. We developed 101-PGC-005 ('005), a macrophage-targeted type 1a dexamethasone prodrug engineered for low-affinity, recycling-compatible uptake via CD206, with intracellular release triggered by acidic endosomes. We evaluated '005 in mechanistic assays, pathogen-diverse preclinical models, three human pharmacokinetic (PK) studies, and an adaptive-design randomized Phase II/III trial in 309 hospitalized patients with moderate COVID-19. In two completed Phase I human studies, a first-in-human dose-escalation and repeated-dose study and a dedicated single/multiple-dose PK and safety study; '005 circulated as intact prodrug with rapid systemic clearance (Tmax ~0.5 h; terminal half-life ~1.9 h), with no measurable free dexamethasone after single dosing and only low, clinically non-significant free dexamethasone after repeated dosing, and intact prodrug recovered unchanged in urine. Morning cortisol and ACTH were preserved after 30 mg once daily for three consecutive days (1.5 times the intended therapeutic dose). A cerebrospinal fluid PK study is evaluating central-compartment penetration. In the Phase II/III trial, powered for non-inferiority, conducted across six sites in India under GCP with Ministry of Health approval and independent DSMB oversight; '005 (20 mg IV daily for 3 days) was superior to dexamethasone (6 mg IV daily for 3 -10 days) on the primary endpoint of time to > a 2-point improvement on the WHO ordinal scale (HR 2.31; 95% CI 1.83-2.93; p < 0.0001; median 3 vs. 4 days). '005 was also superior on viral clearance (HR 1.47; 95% CI 1.17-1.84; p = 0.0001), hospital discharge rate, SpO2; recovery, and fever resolution. Zero patients in the '005 arm received investigator-initiated corticosteroid supplementation despite protocol allowance. All 309 randomized patients completed the study (ITT = per-protocol). Safety profiles were equivalent (TEAEs 54.8% vs 54.5%; p = 0.958), with no Grade 3+ events, SAEs, deaths, or discontinuations in either arm. Mechanistically, '005 delivered dual benefit: acute debulking of inflammatory macrophages and selective depletion of chronically activated pathology-sustaining macrophages, while preserving CXCL10 antiviral signaling and physiologic HPA control. Critically, HPA preservation is not merely a safety feature, it is a core efficacy mechanism: by clearing the pathogenic macrophage burden that was overriding HPA regulation, '005 restores the conditions for endogenous cortisol to resume its pulsatile, demand-responsive anti-inflammatory role across all GR-expressing cells, lymphocytes, endothelial cells, neurons, and newly differentiated macrophages, that '005 itself cannot reach. These findings support regulatory-grade evidence for macrophage-targeted corticosteroid therapy and provide the foundation for further development across acute inflammatory indications (sepsis, viral pneumonia, cytokine-release syndromes) and chronic macrophage-driven diseases (atherosclerosis, metabolic steatohepatitis, neurodegeneration, tumor-associated macrophages).

14.
medRxiv (Medicine) 2026-06-12

Sociodemographic and health correlates of reimbursement authorizations for cannabis for medical purposes in Canadian veterans: A cross-sectional study linking the Life After Services Studies 2019 and Health Administrative Databases

Background Evidence on factors associated with cannabis for medical purposes (CMP) authorizations among Veterans Affairs Canada (VAC) clients remains limited and inconsistent, particularly concerning mental health and posttraumatic stress disorder (PTSD), a leading indication for use. We investigated demographic, clinical and service characteristics associated with VAC authorizations for CMP reimbursement. Method We linked VAC administrative CMP program data with responses from the 2019 Life After Services Studies cross-sectional survey of Regular Force veterans released between 1998 and 2018. Multivariable logistic regressions examined associations between CMP reimbursement (yes/no) and demographic, clinical and well-being factors, with analyses stratified by PTSD status. Results Among 1,289 respondents (weighted n=33,131), 18.4% were authorized for CMP reimbursement. Younger age (

15.
arXiv (CS.CL) 2026-06-15

MedLatentDx: Latent Multi-Agent Communication for Cross-Hospital Rare-Disease Diagnosis

Rare diseases affect over $300$ million patients across more than $7{,}000$ conditions, yet no single hospital encounters enough cases of any one condition for reliable diagnosis. Cross-hospital collaboration could help by allowing a diagnosing institution to use distributed, case-specific diagnostic evidence, but privacy regulations restrict the transmission of identifiable clinical text across institutional boundaries. This setting raises two challenges: existing medical agent systems often rely on textual evidence exchange, while raw latent states such as hidden states and KV caches may still reveal prompt-derived clinical content. We introduce MedLatentDx, a latent multi-agent communication framework in which hospital agents keep private clinical records and retrieved cases local, and send compact latent KV blocks to a host agent for rare-disease diagnosis. MedLatentDx supports two deployment settings: same-backbone hospital agents use latent KV distillation, while hospitals with different LLM backbones use cross-family latent alignment. On CrossRare-Bench, a self-built large-scale rare-disease benchmark with hospital-level partitions, MedLatentDx improves cross-hospital diagnostic performance while reducing reconstructable clinical content relative to raw-latent communication baselines.

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

AI Sovereignty as National Learning Capacity: A Human-Centered Learning Mechanics Viewpoint on France, the United States, and China

arXiv:2606.00729v2 Announce Type: replace Abstract: Artificial intelligence in France is often discussed through separate dimensions such as investment, compute, regulation, employment, sovereignty, and education. This viewpoint paper proposes a unified interpretation: France can be analyzed as a national AI learning system. Building on Human-Centered Learning Mechanics (HCLM), we use HCLM not as a validated econometric model, but as a conceptual and diagnostic lens for interpreting national AI development as a balance between information injection, absorptive capacity, and institutional dissipation. Information injection includes compute, data, talent, research, capital, industrial deployment, and policy experimentation. Institutional dissipation refers to avoidable frictions such as administrative overload, coordination failures, energy constraints, regulatory uncertainty, talent mobility pressures, and weak industrial absorption. Regulation is not treated as mere friction: adaptive governance, trusted data spaces, and safety-oriented standards may increase long-term learning capacity by improving legitimacy, interoperability, and social trust. The central claim is not that a country follows neural-network equations, but that AI sovereignty depends on how effectively it converts distributed information into absorbed, coordinated, and socially legitimate capability. The paper connects HCLM with neural scaling laws, endogenous growth theory, creative destruction, absorptive capacity, and coordination mechanisms. It offers a formal heuristic, policy indicators, illustrative scenarios, and implications for France. The numerical results are diagnostic scenarios, not econometric estimates or official rankings. The proposed viewpoint reframes AI policy as the governance of an open, strategic, non-equilibrium learning system that should be tested with historical and cross-country data.

17.
arXiv (CS.CL) 2026-06-12

A Unifying Lens on Reward Uncertainty in RLHF

Reinforcement learning from human feedback (RLHF) is bottlenecked by reward hacking, where the policy exploits errors in a proxy reward model (RM) and produces high RM scores without genuine quality gains. A natural mitigation is pessimism: lowering rewards in regions where the RM is uncertain. However, standard scalar RMs provide no principled notion of uncertainty. We argue that the right object is a distributional reward model $p(r\mid x,y)$. Under either a Bayesian inference or a KL-distributionally robust optimization (KL-DRO) lens, the KL-regularized RLHF objective admits a closed-form effective reward $\tilde r(x,y) = \pm\beta\log\mathbb{E}_p[e^{\pm r/\beta}]$. The pessimistic branch unifies the prior heuristics for RM ensemble aggregation: mean aggregation, worst-case optimization (WCO), and uncertainty-weighted optimization (UWO) all emerge as limits or truncations of this single expression. This also clarifies the implicit assumptions of each existing rule.

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

SDFLoRA: Selective Decoupled Federated LoRA for Privacy-preserving Fine-tuning with Heterogeneous Clients

arXiv:2601.11219v3 Announce Type: replace-cross Abstract: Federated learning (FL) for large language models (LLMs) has attracted increasing attention as a privacy-preserving approach for adapting models over distributed data, where parameter-efficient methods such as Low-Rank Adaptation (LoRA) are widely adopted to reduce communication and memory costs. However, practical deployments often exhibit rank and data heterogeneity: clients operate under different low-rank budgets and data distributions, making direct aggregation of LoRA updates biased and unstable. Existing approaches either enforce a unified rank or align heterogeneous updates into a single shared subspace, which tends to mix transferable and client-specific directions and consequently undermines personalization. Moreover, under differential privacy (DP), perturbing such structurally mixed updates injects noise into directions that should remain purely local, leading to unnecessary utility degradation. To address these issues, we propose Selective Decoupled Federated LoRA (SDFLoRA), a structure-aware LoRA framework that decouples each client update into a shared component for aggregation and a private component that preserves client-specific semantics. Only the shared component participates in subspace alignment, while the private component remains local and uncommunicated, making the training DP-compatible and stabilizing aggregation under rank heterogeneity. By injecting noise only into the aggregated shareable update, this approach avoids perturbations to local directions and improves the utility-privacy trade-off. Experiments on multiple benchmarks demonstrate that SDFLoRA outperforms federated LoRA baselines and achieves a strong utility-privacy trade-off.

19.
arXiv (CS.CL) 2026-06-16

ACC: Compiling Agent Trajectories for Long-Context Training

Recent development of agents has renewed demand for long-context reasoning capacity of LLMs. However, training LLMs for this capacity requires costly long-document curation or heuristic context synthesis. We observe that agents produce massive trajectories when solving problems, invoking tools and receiving environment observations across many turns. The evidence needed to answer the original question is thus scattered throughout these turns, requiring integration of distant context segments. Nevertheless, standard agent SFT masks tool responses and only trains turn-level tool selection, creating a supervision blind spot where these scattered signals go unused. We propose Agent Context Compilation (ACC), which converts trajectories from search, software engineering, and database querying agents into long-context QA pairs that combine the original question with tool responses and environment observations gathered across multiple turns, training the model to answer directly without tool use. This makes the dependencies between the question and the evidence explicit, enabling direct supervision of long-context reasoning over distant segments without additional annotation. ACC is a simple but effective approach that can be combined with any existing long-context extension or training method, providing scalable supervised fine-tuning data. We validate ACC on long-range dependency modeling tasks through MRCR and GraphWalks, challenging benchmarks requiring cross-turn coreference resolution and graph traversal over extended contexts. Training Qwen3-30B-A3B with ACC achieves 68.3 on MRCR (+18.1) and 77.5 on GraphWalks (+7.6), results comparable to Qwen3-235B-A22B, while preserving general capabilities on GPQA, MMLU-Pro, AIME, and IFEval. Further mechanism analysis reveals that the ACC-trained model exhibits task-adaptive attention restructuring and expert specialization.

20.
arXiv (CS.CV) 2026-06-18

Quantification of Uncertainty with Adversarial Models in Medical Image Segmentation

Reliable pixel-level uncertainty quantification holds the potential to transform clinical workflows by enabling high-fidelity longitudinal monitoring and distinguishing true pathological changes from artifacts. Ideally, these models provide the stability required for critical treatment planning and surgical intervention. However, standard deep learning models often suffer from miscalibration, yielding overconfident predictions that mask underlying vulnerabilities at subtle pathological boundaries. To address this, we propose QUAM-SM, a post-hoc framework using targeted adversarial search to identify "adversarially fragile" pixels. By actively seeking perturbations that expose predictive instability, our method highlights regions where decisions are most vulnerable to being flipped. Importantly, the framework disentangles epistemic uncertainty from aleatoric uncertainty. Experiments on two public datasets with multiple expert annotations demonstrate that QUAM-SM outperforms both standard and recent uncertainty estimation approaches in terms of reliability and boundary sensitivity. Code is available at https://github.com/HanaJebril/quam_sm

21.
arXiv (CS.CL) 2026-06-12

Self-Distillation Zero: Self-Revision Turns Binary Rewards into Dense Supervision

Current post-training methods in verifiable settings fall into two categories. Reinforcement learning (RLVR) relies on binary rewards, which are broadly applicable and powerful, but provide only sparse supervision during training. Distillation provides dense token-level supervision, typically obtained from an external teacher or using high-quality demonstrations. Collecting such supervision can be costly or unavailable. We propose Self-Distillation Zero (SD-Zero), a method that is substantially more training sample-efficient than RL and does not require an external teacher or high-quality demonstrations. SD-Zero trains a single model to play two roles: a Generator, which produces an initial response, and a Reviser, which conditions on that response and its binary reward to produce an improved response. We then perform on-policy self-distillation to distill the reviser into the generator, using the reviser's token distributions conditioned on the generator's response and its reward as supervision. In effect, SD-Zero trains the model to transform binary rewards into dense token-level self-supervision. On math and code reasoning benchmarks with Qwen3-4B-Instruct and Olmo-3-7B-Instruct, SD-Zero improves performance by at least 10% over the base models and outperforms strong baselines, including Rejection Fine-Tuning (RFT), GRPO, and Self-Distillation Fine-Tuning (SDFT), under the same question set and training sample budget. Extensive ablation studies show two novel characteristics of our proposed algorithm: (a) token-level self-localization, where the reviser can identify the key tokens that need to be revised in the generator's response based on reward, and (b) iterative self-evolution, where the improving ability to revise answers can be distilled back into generation performance with regular teacher synchronization. Code: https://github.com/princeton-pli/Self-Distillation-Zero.

22.
arXiv (CS.CL) 2026-06-11

Models That Know How Evaluations Are Designed Score Safer

The validity of AI safety evaluations depends on models behaving consistently across controlled and deployment settings. Prior work has identified test-time contextual cues, such as hypothetical scenarios, as a source of verbalized evaluation awareness and subsequent behavioral shift. In this paper, we investigate a potential explanation of this phenomenon: evaluation meta-knowledge, defined as parametric knowledge about the structural traits that characterize evaluations. Similar to dataset contamination, where benchmark exposure leads to higher performance through memorization, we hypothesize that models trained on texts describing evaluation practices may implicitly learn to recognize and respond to evaluation-like contexts, for instance, through exposure to scientific articles or social media posts about AI benchmarking. To test this, we fine-tune models on synthetic documents describing evaluation traits such as verifiable structures or moral dilemmas. Evaluating this fine-tuned model on six safety benchmarks, we find that it is significantly safer than the base model and control model. This behavioral shift persists even when restricting the analysis to responses lacking explicit verbalization of evaluation awareness. Our results demonstrate that evaluation meta-knowledge may inflate safety benchmark performance, introducing a novel confounder that is independent of explicit memorization or verbalized evaluation awareness, thus, challenging to detect. These findings have important implications for the design and interpretation of AI safety evaluations. Our code and models are available at https://github.com/compass-group-tue/arxiv2026_evaluation_meta_knowledge.

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

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

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

24.
arXiv (quant-ph) 2026-06-17

Manipulation of Topological Corner States via Subchiral Symmetry

arXiv:2606.17975v1 Announce Type: new Abstract: Higher-order topological phases provide robust corner modes, but their use requires controllable creation, isolation, and transfer of individual modes and their superpositions. Here we demonstrate, using the two-dimensional Benalcazar-Bernevig-Hughes model as an example, that subchiral symmetry provides a general control principle for manipulating topological corner modes. The conventional chiral symmetry decomposes into four subchiral symmetries, each associated with one zero-energy corner mode. By selectively breaking these subsymmetries with controlled intercell hoppings, we reduce the fourfold corner-state manifold step by step to single isolated modes. We further design adiabatic protocols that transfer either a single corner state or a superposition of two corner states between selected corners, while preserving the relative phase in the latter case. Both numerical simulations and IBM quantum-processor implementations show that the proposed protocols can be executed with high fidelity, establishing subchiral symmetry as a route to programmable higher-order topological state manipulation.

25.
arXiv (math.PR) 2026-06-18

Denoising Distances in Metric Measure Spaces

arXiv:2606.18301v1 Announce Type: cross Abstract: Recent work studied the problem of finding clusters and denoising pairwise distances from noisy distances of points sampled on a manifold. We study the same problems in more general metric measure spaces under \lowerphiregularity{}. We give an algorithm that extracts large localized clusters around every sampled point and uses them to denoise distances to any fixed accuracy, with near-linear running time in the dense fixed-accuracy regime. We also show how to achieve much higher accuracy with a non-efficient algorithm. This suggests that unlike the Riemannian case, denoising to higher accuracy in more general metric spaces has a statistical-computational gap.