探索全球前沿学术脉络

01.

arXiv (CS.CL) 2026-06-17 DOI: arXiv:2606.17162

MemSlides: A Hierarchical Memory Driven Agent Framework for Personalized Slide Generation with Multi-turn Local Revision

作者:

Ye Jin↗Yangyang Xu↗Jun Zhu↗Yibo Yang↗

Personalized presentation generation requires more than conditioning on a current prompt or template: agents must preserve stable user preferences across tasks, retain newly introduced preferences and constraints during multi-turn revision, and carry out local edits reliably. We propose MemSlides, a hierarchical memory framework for personalized presentation agents that separates long-term memory from working memory and further divides long-term memory into user profile memory and tool memory. User profile memory stores intent-conditioned profiles for round-0 personalization, working memory carries active preferences and session constraints across revision rounds, and tool memory stores reusable execution experience for reliable localized editing. MemSlides pairs this memory design with scoped slide-local revision, so targeted updates act on the smallest affected region instead of repeatedly regenerating the full deck. In controlled experiments, user profile memory improves persona-alignment judgments on a multi-persona, multi-intent profile bank, tool-memory injection improves closed-loop modify behavior in diagnostic matched-pair settings, and qualitative cases illustrate working memory's ability to carryover preferences. Taken together, these results suggest that effective personalization in presentation authoring depends on separating persistent user profiles, session-level working memory, and reusable execution experience across generation and localized revision.

阅读与讨论 → 访问原文 →

02.

arXiv (CS.CL) 2026-06-15 DOI: arXiv:2606.14243

Decoupled Mixture-of-Experts for Parametric Knowledge Injection

作者:

Baoqing Yue↗Weihang Su↗Qingyao Ai↗Yichen Tang↗Changyue Wang↗Jiacheng Kang↗Jingtao Zhan↗Yiqun Liu↗

Knowledge injection aims to equip large language models (LLMs) with external, domain-specific, or time-sensitive knowledge. Existing approaches typically face a trade-off between flexibility and integration: retrieval-augmented generation keeps knowledge outside the model but only provides prompt-level augmentation, whereas post-training based methods encode new knowledge into shared parameters but may introduce catastrophic forgetting, knowledge conflict, and costly updates. In this paper, we propose Decoupled Mixture-of-Experts (DMoE), a modular architecture for parametric knowledge injection that decouples both experts and the router from the base model. DMoE converts external knowledge corpora into independently updatable expert modules and uses a lightweight uncertainty-aware router to activate relevant experts only when the base model lacks sufficient knowledge during generation. To support efficient auto-regressive inference, DMoE attaches experts only to the final-layer feed-forward network, preserving KV-cache reuse while enabling parameter-level knowledge augmentation. Experiments on knowledge-intensive benchmarks show that DMoE consistently improves answer quality over retrieval and adapter-based baselines.

阅读与讨论 → 访问原文 →

03.

arXiv (CS.LG) 2026-06-15 DOI: arXiv:2606.14668

When to Write and When to Suppress: Route-Specialized Dual Adapters for Memory-Assisted Knowledge Editing

作者:

Yining Huang↗

arXiv:2606.14668v1 Announce Type: new Abstract: Knowledge editing systems must update selected facts while preserving nearby but irrelevant behavior. This paper studies this problem in a memory-assisted setting where an edit memory is retrieved at inference time and a parameter-efficient adapter corrects the model's object preference. We argue that the central design question is not only how to write an edit, but also when to suppress it. We introduce \method{}, a route-specialized dual-adapter editor. A relevance router first decides whether a prompt should receive an edit memory. Routed prompts use an edit adapter trained to prefer the new object over the original object; unrouted non-direct prompts use a separate locality adapter trained to preserve or restore the original-object preference. We evaluate \method{} on three 1,000-case protocols, \cf{}, \zsre{}, and \mquake{}, under the same memory protocol and two 7B/8B base models. On Llama-3.1-8B-Instruct, \method{} obtains the best overall probability-preference accuracy on all three benchmarks: 0.8180 on \cf{}, 0.8946 on \zsre{}, and 0.9922 on \mquake{}. The same trend holds on Qwen3-8B. Router ablations show that the relevant memory boundary differs across datasets: a lexical neural router is safest on \cf{}, while BGE embedding routing is better on \zsre{} and \mquake{}. Component and module ablations show that the gain mainly comes from separating edit injection from off-route suppression rather than from simply increasing LoRA capacity.

阅读与讨论 → 访问原文 →

04.

arXiv (CS.CV) 2026-06-17 DOI: arXiv:2606.17557

Universal Image Restoration via Internalized Chain-of-Thought Reasoning

作者:

Yu Guo↗Zhengru Fang↗Shengfeng He↗Senkang Hu↗Yihang Tao↗Phone Lin↗Yuguang Fang↗

Image restoration seeks to recover high-quality images from degraded inputs but becomes highly ill-posed under complex, mixed degradations. While unified all-in-one models are common, their performance declines as degradation complexity increases. Recent works adopt Chain-of-Thought (CoT) reasoning for multi-round restoration using specialized modules. However, this approach faces two key limitations: (i) increased computational cost due to multi-step processing, and (ii) weak modeling of interactions between degradations during stepwise inference. We introduce CoTIR, a universal image restoration framework that internalizes CoT reasoning within a single model. Concretely, we view image restoration as a specialized subtask of image editing, which implies that a large-scale pre-trained editing model provides a more favorable optimization starting point. Building on this, we fine-tune the model for restoration and further encode structured CoT-style reasoning into the learning objective via a differentiable formulation inspired by Lagrangian optimization, enabling holistic restoration without chaining specialized restorers. To facilitate training and evaluation, we further present CoTIR-Bench, a large-scale benchmark comprising 5.2 million samples with CoT-style reasoning traces. Extensive experiments on CoTIR-Bench and broad real composite degradation scenes show that CoTIR achieves stronger perceptual quality and more competitive fidelity than both all-in-one models and multi-round restoration methods. The source code is available at https://github.com/gy65896/CoTIR.

阅读与讨论 → 访问原文 →

05.

arXiv (quant-ph) 2026-06-19 DOI: arXiv:2603.18399

Accelerated Rydberg electromagnetically induced transparency quantum memory via shortcuts to adiabaticity

作者:

Y. Wei↗Changcheng Li↗Y. M. Liu↗Yuechun Jiao↗Weibin Li↗X. Q. Shao↗

arXiv:2603.18399v2 Announce Type: replace Abstract: Electromagnetically induced transparency (EIT) enables coherent light-matter storage, forming the basis of photonic quantum memories that are essential for scalable quantum networks and distributed quantum computing. However, accelerating the storage process violates the adiabatic condition, resulting in the excitation of the lossy intermediate state and a reduction in writing efficiency. We propose and numerically investigate a high-speed, high-fidelity quantum storage scheme by incorporating a shortcut-to-adiabaticity (STA) technique based on counter-diabatic (CD) driving. By introducing a precisely engineered auxiliary field into a conventional EIT system, our protocol significantly shortens the writing time beyond the conventional adiabatic limit while effectively suppressing the transient population of the lossy intermediate state. Furthermore, our scheme demonstrates strong flexibility in pulse design, remaining effective across different temporal profiles of both the control and signal fields. It also exhibits robustness against imperfections in the CD drive. Even with imperfect single-photon writing and non-ideal Rydberg blockade, the scheme retains clear advantages, maintaining high storage performance and overcoming the intrinsic speed-fidelity trade-off of traditional EIT protocols. These features pave the way for fast and robust quantum devices suitable for high-throughput quantum repeaters and advanced quantum information processing.

阅读与讨论 → 访问原文 →

06.

arXiv (CS.CL) 2026-06-12 DOI: arXiv:2606.13572

ArogyaSutra: A Multi-Agent Framework for Multimodal Medical Reasoning in Indic Languages

作者:

Tanmoy Kanti Halder↗Akash Ghosh↗Subhadip Baidya↗Arijit Roy↗Sriparna Saha↗

Multimodal Large Language Models (MLLMs) have shown promising reasoning capabilities in general domains, yet their performance remains limited in specialized settings such as healthcare, especially in multilingual and low-resource scenarios. This gap is critical in regions like rural India, where patients often express complex medical queries in native Indic languages and rely on multimodal inputs such as medical images. Existing English-centric MLLMs struggle to support such use cases, limiting equitable access to AI-driven healthcare assistance. To address this challenge, we introduce ArogyaBodha, a large-scale multilingual multimodal medical question-answer dataset constructed from eight heterogeneous sources, covering 31 body systems, six imaging modalities, and 21 clinical domains across English and seven major Indian languages. We further propose ArogyaSutra, an actor-critic-based multi-agent framework that integrates tool grounding with dual-memory mechanisms for step-wise, reasoning-aware decision making, and uses stored actor-critic simulation trajectories for distillation. Experiments show that our dataset and framework improve multilingual medical reasoning accuracy across all Indic languages, with ablations validating the contribution of each component. The source code and dataset are available at: https://iitp-cse.github.io/ ArogyaSutra/

阅读与讨论 → 访问原文 →

07.

medRxiv (Medicine) 2026-06-15 DOI: HASH:b239d8187fe84995897c4fd55fbe9868

Quantitative Gait Categorization in Parkinson's Disease with and without Freezing of Gait

作者:

Donovan↗Tripathi↗Chu↗Bernhard↗Factor↗McKay↗J. L↗Esper↗

Background: Freezing of gait (FOG) is a disabling and often underrecognized feature of Parkinsons disease (PD). Objective gait analysis may improve characterization of this motor symptom. Objective: To compare quantitative 3D gait parameters in PD with FOG (PDF) and PD without FOG (PDNF) in a routine clinical cohort. Methods: We retrospectively analyzed a sequential sample of 180 patients with PD referred for motion analysis between 2020 and 2024. All patients underwent 3D motion capture in the off-medication state. Eighteen gait outcomes spanning pace, rhythm, postural control, variability, and asymmetry domains were derived from steady-state walking tasks. FOG status was determined using physician documentation and Movement Disorder Society Unified Parkinsons Disease Rating Scale (MDS-UPDRS) items. Group differences between PDF (n=99) and PDNF (n=81) were evaluated using independent samples t-tests, with outcomes adjusted for disease duration and corrected for multiple comparisons. A secondary analysis among PDF compared those in Hoehn and Yahr (H&Y) stage [≥]III to those in H&Y [≤]II. Results: PDF had longer disease duration, higher OFF MDS-UPDRS III scores, and higher Hoehn and Yahr stage than PDNF but were similar in age and sex. After adjusting for disease duration and multiplicity, PDF demonstrated reduced step length, stride length, and forward velocity, and greater cadence variability, while most postural control, and asymmetry measures were comparable between groups. Among PDF, advanced H&Y stage was associated with impaired pace and rhythm, similar to previous reports among PD in general. Conclusion: In this large, sequential, clinically referred cohort, FOG was associated with more advanced PD and specific impairments in pace and gait variability. These findings support comprehensive 3D gait analysis as an objective tool to better delineate FOG-related gait abnormalities and identify features that may predict FOG, informing targeted interventions.

阅读与讨论 → 访问原文 →

08.

arXiv (CS.LG) 2026-06-16 DOI: arXiv:2606.16379

Scalable and Interpretable Representation Alignment with Ordinal Similarity

作者:

Diogo Soares↗Pankhil Gawade↗Andrea Dittadi↗Ewa Szczurek↗

arXiv:2606.16379v1 Announce Type: new Abstract: Evaluating representation similarity is fundamental to representation learning. However, existing metrics suffer from significant limitations: they lack interpretability due to shifting baselines, lack robustness to outliers, and are computationally intractable for large datasets, forcing reliance on heuristic approximations. To address this, we develop an ordinal-similarity framework, instantiated by the Triplet (TSI) and Quadruplet (QSI) Similarity Indices, which measure alignment by quantifying the consistency of ordinal relationships. We theoretically demonstrate this formulation is inherently interpretable, robust to outliers, and computationally efficient. Finally, we establish a formal equivalence between TSI and local neighborhood alignment, measured by Mutual Nearest Neighbors. Empirically, we validate these properties and show that ordinal similarity offers a scalable approach to measuring alignment, enabling practitioners to better understand and design representations.

阅读与讨论 → 访问原文 →

09.

arXiv (CS.LG) 2026-06-12 DOI: arXiv:2606.13614

Majority-of-Three is Optimal

作者:

Divit Rawal↗Nikita Zhivotovskiy↗

arXiv:2606.13614v1 Announce Type: cross Abstract: We give a short proof that the majority vote of three independent consistent classifiers is an optimal learner in the realizable PAC setting. This proves optimality for the simplest voting scheme, while simplifying both the algorithmic structure and the probabilistic analysis of previous voting learners, including the algorithm of S. Hanneke and the analysis of bagging by K. Green Larsen.

阅读与讨论 → 访问原文 →

10.

arXiv (CS.AI) 2026-06-16 DOI: arXiv:2606.14971

FastMix: Fast Data Mixture Optimization via Gradient Descent

作者:

Haoru Tan↗Sitong Wu↗Yanfeng Chen↗Jun Xia↗Ruobing Xie↗Bin Xia↗Xingwu Sun↗Xiaojuan Qi↗

arXiv:2606.14971v1 Announce Type: cross Abstract: While large and diverse datasets have driven recent advances in large models, identifying the optimal data mixture for pre-training and post-training remains a significant open problem. We address this challenge with FASTMIX, a novel framework that automates data mixture discovery while training only a single proxy model. Instead of relying on predefined heuristics or resource-intensive simulations, FASTMIX jointly optimizes mixture coefficients and model parameters, substantially improving efficiency and scalability over prior approaches. At the core of FASTMIX is a reformulation of mixture selection as a bilevel optimization problem. Under this reformulation, we show that optimizing mixture ratios is mathematically equivalent to assigning per-source loss weights under uniform source sampling. This embeds the mixture coefficients directly into the differentiable iterative optimization objective, enabling efficient, gradient-based optimization of both mixture and model. To solve the optimization problem, FASTMIX implements an approximate iterative optimization procedure, alternating between (i) updating model parameters on data sampled according to current mixture ratios (inner loop) and (ii) updating mixture ratios based on validation feedback (outer loop). Across pre- and post-training, FASTMIX outperforms baselines while drastically reducing search cost. Code (https://github.com/hrtan/fastmix)

阅读与讨论 → 访问原文 →

11.

arXiv (CS.CV) 2026-06-11 DOI: arXiv:2606.11930

Frozen Multimodal Embeddings for Personality and Cognitive Ability Assessment in Asynchronous Video Interviews

作者:

Kuo-En Hung↗Hung-Yue Suen↗Shih-Ching Yeh↗Hsiang-Wen Wang↗

Predicting psychological traits from asynchronous video interviews (AVIs) is a challenging multimodal learning problem because labeled datasets are limited while each response contains high-dimensional visual, acoustic, and verbal signals. This paper presents our solution for the ACM Multimedia AVI Challenge 2026, which evaluates two tasks: Track~1 predicts self-reported HEXACO personality traits from personality-related interview responses, and Track~2 classifies cognitive ability levels from structured AVI responses. We treat the problem as a small-sample representation learning task. Instead of fine-tuning large pretrained models, we use frozen multimodal encoders, including CLIP for visual features, Whisper for acoustic features and transcripts, and RoBERTa, E5, and DeBERTaV3 for textual representations, followed by low-capacity downstream models. For Track~1, our trait-specific regression and late-fusion system achieves an average validation MSE of 0.2696, improving over the official baseline of 0.3334. Ablation results show a three-step improvement from a global model (0.3189), to per-trait modeling (0.2871), to per-trait late fusion (0.2696), corresponding to a 19.1\% relative MSE reduction over the official baseline. For Track~2, a compact subject-attribute baseline reaches 0.5781 accuracy, while our multimodal ensemble reaches 0.5313, both above the official baseline of 0.4062. We interpret this result as evidence of possible subject-attribute shortcuts in the validation split rather than robust cognitive inference from AVI content. Overall, our findings suggest that AVI-based psychological assessment benefits from trait-specific multimodal modeling, but cognitive ability prediction requires careful control of dataset shortcuts.

阅读与讨论 → 访问原文 →

12.

arXiv (CS.AI) 2026-06-12 DOI: arXiv:2512.15133

HD-Prot: A Protein Language Model for Joint Sequence-Structure Modeling with Continuous Structure Tokens

作者:

Yi Zhou↗Haohao Qu↗Yunqing Liu↗Shanru Lin↗Le Song↗Wenqi Fan↗

arXiv:2512.15133v3 Announce Type: replace-cross Abstract: Proteins inherently possess a consistent sequence-structure duality. The abundance of protein sequence data, which can be readily represented as discrete tokens, has driven fruitful developments in protein language models (pLMs). A key remaining challenge, however, is how to effectively integrate continuous structural knowledge into pLMs. Current methods often discretize protein structures to accommodate the language modeling framework, which inevitably results in the loss of fine-grained information and limits the performance potential of multimodal pLMs. In this paper, we argue that such concerns can be circumvented: a sequence-based pLM can be extended to incorporate the structure modality through continuous tokens, i.e., high-fidelity protein structure latents that avoid vector quantization. Specifically, we propose a hybrid diffusion protein language model, HD-Prot, which embeds a continuous-valued diffusion head atop a discrete pLM, enabling seamless operation with both discrete and continuous tokens for joint sequence-structure modeling. It captures inter-token dependencies across modalities through a unified absorbing diffusion process, and estimates per-token distributions via categorical prediction for sequences and continuous diffusion for structures. Extensive results demonstrate that HD-Prot achieves competitive performance in unconditional sequence-structure co-generation, motif-scaffolding, protein structure prediction, and inverse folding tasks. Furthermore, our method can perform on par with state-of-the-art multimodal pLMs, despite being developed under limited computational resources (i.e., less than one-tenth the budget for modality extension fine-tuning). It highlights the viability of simultaneously estimating categorical and continuous distributions within a unified language model architecture, offering a promising alternative direction for multimodal pLMs.

阅读与讨论 → 访问原文 →

13.

arXiv (CS.LG) 2026-06-16 DOI: arXiv:2606.16304

pFedUL: Layer-Aware Federated Unlearning for Personalized Federated Learning

作者:

Zhuodong Liu↗Xiangyu Li↗Zhihao Zhang↗

arXiv:2606.16304v1 Announce Type: new Abstract: Federated unlearning (FU) enables the removal of specific data contributions from federated learning (FL) models to comply with regulations such as the General Data Protection Regulation (GDPR). However, most existing FU methods are designed for the FedAvg paradigm, where all clients share a single global model. In practice, personalized federated learning (pFL) methods such as FedPer, FedRep, Ditto, and FedBN have become widely adopted due to their superior handling of non-IID data. These methods decompose the model into shared global layers and client-specific personalized layers, fundamentally altering the semantics of unlearning, yet this setting has received little attention. We formalize FU under the pFL paradigm, identifying a tension between unlearning completeness on shared layers and personalization preservation for remaining clients. We then propose pFedUL, a layer-aware selective unlearning framework comprising three components: (1) gradient-based layer-wise contribution attribution that separately quantifies the target client's influence on shared and personalized parameters, (2) adaptive selective unlearning that applies differentiated forgetting strategies across layer types, and (3) a lightweight recalibration protocol enabling remaining clients to restore personalization with minimal overhead. We further introduce two new metrics, Personalization Preservation Score (PPS) and Cross-client Fairness Index (CFI), to evaluate pFL-specific unlearning quality. Experiments on CIFAR-10, CIFAR-100, and FEMNIST under varying non-IID settings indicate that pFedUL achieves unlearning effectiveness comparable to full retraining while maintaining an average of 97.3\% personalized accuracy for remaining clients. Compared with six state-of-the-art FU methods adapted to the pFL setting, pFedUL consistently achieves superior personalization preservation.

阅读与讨论 → 访问原文 →

14.

arXiv (CS.AI) 2026-06-17 DOI: arXiv:2606.18119

First Proof Second Batch

作者:

Mohammed Abouzaid↗Nikhil Srivastava↗Rachel Ward↗Lauren Williams↗

arXiv:2606.18119v1 Announce Type: new Abstract: To assess the ability of current AI systems to correctly solve research-level mathematics problems, we tested several AI systems on a set of ten problems in a broad range of mathematical fields; these problems arose naturally in the research process of the contributors. This document includes the problems, our methodology, and the results of our testing. We provide links to supplementary documents including the human solutions, the AI-generated solutions, and the referee reports and logs for the AI-generated solutions. The ten problems were contributed by the following mathematicians: (1) Dariusz Kaloci\'nski and Theodore A. Slaman, (2) Richard Schwartz, (3) Aleksa Milojevic and Benny Sudakov, (4) Larry Guth, (5) Oleg Butkovsky, Jonathan Mattingly, and Lorenzo Zambotti, (6) Joshua Evan Greene and Duncan McCoy, (7) Sucharit Sarkar, (8) Sam Payne and Jidong (Jayden) Wang, (9) Sylvie Corteel and John Lentfer, (10) Srivatsav Kunnawalkam Elayavalli.

阅读与讨论 → 访问原文 →

15.

arXiv (CS.CL) 2026-06-18 DOI: arXiv:2606.18831

Beyond Reward Engineering: A Data Recipe for Long-Context Reinforcement Learning

作者:

Xiaoyue Xu↗Sikui Zhang↗Xiaorong Wang↗Xu Han↗Chaojun Xiao↗

Long-context reasoning is an essential capability for large language models, particularly when they are deployed as autonomous agents that must reason over lengthy trajectories. Reinforcement learning (RL) has recently emerged as a dominant paradigm for improving this ability, yet existing work largely focuses on reward engineering while diverse training data remains scarce. We revisit this problem from a data-centric perspective and show that a simple yet effective data recipe alone, paired with a minimal outcome-based GRPO setup, suffices to substantially improve long-context reasoning. Our recipe targets three complementary task families – retrieval, multi-evidence synthesis, and reasoning – for which we construct and curate eight datasets totaling ~14K examples. Experiments on three models (Qwen3-4B/8B/30B-A3B) yield average gains of +7.2/+3.2/+6.4 points across seven long-context benchmarks, surpassing prior RL training sets. We further demonstrate that these gains transfer to agentic tasks, where continuing RL training on an agent-tuned model with our data recipe improves GAIA by +4.8 and BrowseComp by +7.0 points. We will release our datasets to facilitate future research.

阅读与讨论 → 访问原文 →

16.

arXiv (CS.LG) 2026-06-19 DOI: arXiv:2606.20082

Beyond Averaging in John Ellipsoid Approximation: High-Accuracy Algorithms in the Leverage-Score Model

作者:

Xiaoyu Li↗Junwei Yu↗Jiaojiao Jiang↗Junbin Gao↗Andi Han↗

arXiv:2606.20082v1 Announce Type: cross Abstract: The John ellipsoid of a symmetric polytope $P=\{\mathbf{x}\in\mathbb{R}^d:\|\mathbf{A}\mathbf{x}\|_\infty\le1\}$, $\mathbf{A}\in\mathbb{R}^{n\times d}$, is computed by a long line of leverage-score algorithms, from Cohen, Cousins, Lee and Yang (COLT 2019) to its successors [WY24, CLS+25], all reaching a $(1+\varepsilon)$-approximation in $\Theta(\varepsilon^{-1}\log(n/d))$ iterations. We separate this complexity into three costs the modern line conflates (certification, identification, and accuracy) and locate the historical $\varepsilon^{-1}$ in the first alone. In the equivalent D-optimal-design form $\min_{\mathbf{p}\in\Delta_n}-\log\det(\sum_i p_i\mathbf{a}_i\mathbf{a}_i^\top)$, the leverage-score oracle is exactly the first-order oracle and the $(1+\varepsilon)$-John guarantee the Frank-Wolfe gap $g(\mathbf{p})\le\varepsilon d$; through this dictionary the costs come apart. The $\varepsilon^{-1}$ is a certification artifact: the uniform average of the iterates, the certificate used throughout the line, has gap exactly $\Theta(1/T)$, however cheap each iteration is made. Pointed instead at the last iterate the same oracle is fast: a warm-started accelerated method reaches the guarantee in $C(\mathbf{A})+O(\sqrt{\kappa}\log(1/\varepsilon))$ queries after an $\varepsilon$-independent setup $C(\mathbf{A})$, and once the optimal face is identified the facial problem is an unconstrained self-concordant minimization whose Hessian the oracle recovers exactly, so damped Newton needs only $O(\log\log(1/\varepsilon))$ steps, for a total of $C(\mathbf{A})+O(d^2\log\log(1/\varepsilon))$ queries. The accuracy dependence is thus doubly logarithmic after an $\varepsilon$-independent, condition-dependent setup; the open problem is the remaining identification cost (a condition-free bound on reaching the optimal face) and lower bounds. Accuracy is not the obstruction.

阅读与讨论 → 访问原文 →

17.

arXiv (CS.LG) 2026-06-15 DOI: arXiv:2606.14536

Provably Safe, Yet Scalable Reinforcement Learning

作者:

Kai S. Yun↗Zeyang Li↗Navid Azizan↗

arXiv:2606.14536v1 Announce Type: new Abstract: Safe reinforcement learning (RL) aims to learn policies that optimize rewards while satisfying constraints. Predominant approaches rely on soft-constrained policy optimization, which has achieved empirical success but does not provide formal safety guarantees for the learned policy. In contrast, methods with strict guarantees typically rely on explicit certificate functions, whose construction requires the direct synthesis and verification of control-invariant sets, a process that scales poorly with state dimension and often yields overly conservative behavior. In this paper, we present the Provably Safe, yet Scalable RL (PS2-RL) framework, a novel two-phase architecture for learning provably safe policies in a scalable manner, designed to overcome the key bottlenecks of prior methods. Rather than explicitly computing invariant sets, PS2-RL leverages a learned backup policy to forward-integrate the system dynamics, generating an implicit control-invariant set online. In the first phase, the backup policy is trained with our proposed safe-arrival value function, which characterizes the optimal backup policy for invariant-set construction. In the second phase, an RL policy is trained end-to-end through a differentiable projection layer that strictly enforces the safety guarantees induced by the learned backup policy. By maximizing the volume of the implicit control-invariant set in the first phase, the resulting PS2 policy from the second phase is performant and scalable, while maintaining provable safety. Crucially, PS2-RL imposes no restrictions on the underlying RL algorithm and can be plugged into any existing training pipeline. We establish theoretical guarantees for the proposed framework and evaluate it on robotic control tasks with state dimensions up to 10, a regime in which prior provably safe RL methods struggle or become impractical.

阅读与讨论 → 访问原文 →

18.

arXiv (CS.AI) 2026-06-15 DOI: arXiv:2606.13767

Beyond LoRA: Is Sparsity-Induced Adaptation Better?

作者:

Elijah Cadenhead↗Cristian McGee↗Xin Li↗El Houcine Bergou↗Aritra Dutta↗

arXiv:2606.13767v1 Announce Type: cross Abstract: Low-rank adaptation (LoRA) and its variants provide a memory- and compute-efficient alternative to full fine-tuning of pre-trained models. However, questions remain about the comparative generalizability of these approaches and how the structural restrictions on low-rank updates preserve effective adaptation performance. We present a historical framing, covering the past (full fine-tuning and original LoRA), the present (different variants of LoRA), and propose simpler, cheaper, parameter-efficient extensions by inducing sparsity within existing LoRA variants: Cheap LoRA (cLA), training a single low-rank factor with the other fixed (deterministically or, in its randomized variant, stochastically), and the chained circulant variant, ${c}^3$LA. We frame cLA as a structured instance of asymmetric LoRA, serving as a controlled column-subspace restriction of full fine-tuning. We derive information-theoretic generalization error bounds for these variants, marking one of the first endeavors in this area. Empirically, we evaluate 11 fine-tuning methods across 10 pre-trained models and 14 datasets, analyzing the fine-tuned models' performance and generalization using tools such as loss landscapes and spectral analysis. Despite the sensitivity of fine-tuned models to the pre-trained model, datasets, and other factors, our study suggests that restricting LoRA-based PEFT methods' adaptation to a sparse, structured column space remains competitive across tasks with their parameter-matched baselines while reducing up to 10% training time and peak GPU memory up to 15%, even with a naïve, non-optimized, sparse implementation. Our theoretical and empirical generalization measures provide a more consistent and principled approach to their cost-effective adaptation than commonly used analytical tools. Overview and code are available at: https://elicaden.github.io/Beyond_LoRA/.

阅读与讨论 → 访问原文 →

19.

arXiv (CS.CV) 2026-06-16 DOI: arXiv:2606.16168

Fi-Gaussian: Frequency-Aware Implicit Gaussian Splatting for Single Image Dehazing

作者:

Yuhan Chen↗Ying Fang↗Guofa Li↗Wenxuan Yu↗Yicui Shi↗Kunyang Huang↗Wenbo Chu↗Keqiang Li↗

Single image dehazing continues to be hindered by the loss of high-frequency details and the difficulty of accurate physical scattering modeling. To address these issues, we propose Fi-Gaussian, a frequency-aware implicit Gaussian splatting network for single image dehazing. Unlike explicit rendering methods that rely on 3D point clouds, our method employs implicit Gaussian splatting to adaptively model the underlying distribution of clear images as a continuous representation in 2D feature space. The core of the network is a frequency-aware implicit Gaussian splatting module, which decouples low-frequency structural information and high-frequency texture information in the frequency domain and then performs adaptive Gaussian aggregation with complex-valued weights to recover fine details. In addition, a physics-driven scattering renormalization mechanism is introduced to estimate the transmission map and atmospheric light under the guidance of implicit Gaussian priors. Extensive experiments on multiple benchmark datasets demonstrate that Fi-Gaussian achieves state-of-the-art quantitative performance and produces visually superior dehazed results, validating the effectiveness of implicit Gaussian splatting for low-level vision tasks.

阅读与讨论 → 访问原文 →

20.

arXiv (CS.LG) 2026-06-11 DOI: arXiv:2506.01396

Mitigating Disparate Impact of Differentially Private Learning through Bounded Adaptive Clipping

作者:

Linzh Zhao↗Aki Rehn↗Mikko A. Heikkil\"a↗Razane Tajeddine↗Antti Honkela↗

arXiv:2506.01396v2 Announce Type: replace Abstract: Differential privacy (DP) has become an essential framework for privacy-preserving machine learning. Existing DP learning methods, however, often have disparate impacts on model predictions, e.g., for minority groups. Gradient clipping, which is often used in DP learning, can suppress larger gradients from challenging samples. We show that this problem is amplified by adaptive clipping, which will often shrink the clipping bound to tiny values to match a well-fitting majority, while significantly reducing the accuracy for others. We propose bounded adaptive clipping, which introduces a tunable lower bound to prevent excessive gradient suppression. Our method improves worst-class accuracy by over 10 percentage points on Skewed and Fashion MNIST compared to unbounded adaptive clipping, 7 points compared to Automatic clipping, and 5 points compared to constant clipping. The code is available at https://github.com/TrustworthyMLHelsinki/adaptive-clipping-fairness.

阅读与讨论 → 访问原文 →

21.

arXiv (quant-ph) 2026-06-11 DOI: arXiv:2606.11947

Controlled ion-ion interactions and cavity-enhanced emission of a coherent dinuclear Eu$^{3+}$ complex

作者:

Evgenij Vasilenko↗Vishnu Unni Chorakkunnath↗Barbora Brachnakova↗Nicholas Lester Jobbitt↗Senthil Kumar Kuppusamy↗David Hunger↗Mario Ruben↗

arXiv:2606.11947v1 Announce Type: new Abstract: Molecular rare-earth-ion complexes offer unique opportunities for quantum technologies by combining the intrinsic coherence properties of rare-earth ions with chemically tunable molecular environments. A crucial capability is the realization of multi-qubit architectures with defined qubit couplings to enable two-qubit quantum gates. Here, we investigate the optical coherence properties and excitation-induced interactions of two Eu$^{3+}$-based molecular complexes, comparing a mononuclear reference system with a dinuclear analogue in which two Eu$^{3+}$ ions are positioned at a well-defined intramolecular distance of about 7 Angstrom. Using cryogenic ensemble spectroscopy, including spectral hole burning, free-induction decay, and photon echo measurements at temperatures down to 100 mK, we demonstrate long optical coherence times $T_{2,o}$ of up to 9 $\mu$s. As a key step toward scalable multi-qubit architectures, a control-target sequence was implemented to probe conditional ion-ion interactions, revealing a stronger interaction-induced dephasing in the dinuclear complex. Finally, we show the integration of the dinuclear complex into a fiber-based optical microcavity, and observe an 380-fold emission enhancement of the $\mathrm{}^5\mathrm{D}_0\rightarrow\mathrm{}^7\mathrm{F}_0$ transition. Together, these results position molecular rare-earth complexes as versatile and chemically tunable building blocks for scalable quantum technologies.

阅读与讨论 → 访问原文 →

22.

arXiv (CS.AI) 2026-06-18 DOI: arXiv:2606.18812

Reinforcement Learning Foundation Models Should Already Be A Thing

作者:

Abdelrahman Zighem↗Jill-J\^enn Vie↗

arXiv:2606.18812v1 Announce Type: cross Abstract: Foundation models for language and vision are powered by internet-scale data, while structured domains (tabular prediction, time-series forecasting, graph learning, reinforcement learning) are not. The substitute is synthetic data, which shifts the burden from collection to prior design. Such priors already exist for many structured tasks: TabPFN and its successors solve tabular classification with a transformer pretrained on a synthetic Bayesian prior. We make two points. First, reinforcement learning is the conspicuous gap: sampling a synthetic MDP is as feasible as sampling a synthetic tabular dataset, yet no in-context RL work treats prior design as a primary objective. Second, MDPs admit a fixed-size sufficient statistic, independent of the episodes observed and tabular in shape, which makes them directly amenable to the attention-based architectures used for tabular foundation models, with a policy head replacing the supervised target. Together these define the agenda for an RL foundation model. As a proof of concept, we train one model entirely on synthetic MDPs and show that, with no task-specific tuning, it solves held-out tabular benchmarks in context, both online and offline: online, in far fewer episodes than UCB-VI and tabular Q-learning, and offline, competitively with VI-LCB.

阅读与讨论 → 访问原文 →

23.

arXiv (quant-ph) 2026-06-12 DOI: arXiv:2505.18367

Improving Variational Counterdiabatic Driving with Weighted Actions and Computer Algebra

作者:

Naruo Ohga↗Takuya Hatomura↗

arXiv:2505.18367v4 Announce Type: replace Abstract: Variational counterdiabatic (CD) driving is a disciplined and widely used method to robustly control quantum many-body systems by mimicking adiabatic processes with high fidelity and reduced duration. Central to this technique is a universal structure of the adiabatic gauge potential (AGP) over a parameterized Hamiltonian. Here, we reveal that introducing a new degree of freedom into the theory of the AGP can significantly improve variational CD driving. Specifically, we find that the algebraic characterization of the AGP is not unique, and we exploit this nonuniqueness to develop the weighted variational method for deriving a refined driving protocol. This approach extends the conventional method in two aspects: it assigns customized weights to matrix elements relevant to specific problems, and it effectively incorporates nonlocal information into local driving coefficients. We also develop an efficient numerical algorithm to compute the refined driving protocol using computer algebra. Our framework is broadly applicable and, in principle, it can replace any previous use of variational CD driving. We demonstrate its practicality by applying it to adiabatic evolution along the ground state of a parameterized Hamiltonian. This proposal outperforms the conventional method in terms of fidelity, as confirmed by extensive numerical simulations on quantum Ising models.

阅读与讨论 → 访问原文 →

24.

arXiv (CS.CV) 2026-06-17 DOI: arXiv:2510.11709

Adversarial Attacks Leverage Interference Between Features in Superposition

作者:

Edward Stevinson↗Lucas Prieto↗Melih Barsbey↗Tolga Birdal↗

Why do adversarial examples exist, and why do they transfer between models? Existing explanations appeal to high-dimensional geometry, non-robust patterns in the input, and decision boundary structure, but none provides a representation-level mechanism that explains why specific perturbations succeed and why attacks transfer between models. In this paper, we show that adversarial vulnerability can stem from efficient information encoding in neural networks. Specifically, vulnerability can arise from superposition - the phenomenon where networks represent more concepts than they have dimensions, forcing non-orthogonal representation and thus interference. This interference causes perturbations targeting one representation to affect others, creating vulnerabilities determined by interference patterns. In synthetic settings with precisely controlled superposition, we establish that superposition suffices to create adversarial vulnerability. The resulting attacks are predictable: PGD-discovered perturbations align with theoretically optimal perturbations derived from the interference geometry. Models trained on similar data develop similar interference patterns, explaining attack transferability. We then show that successful attacks on image classifiers exhibit the structure predicted by our proposed mechanism. These findings reveal that adversarial vulnerability can be a byproduct of networks' representational compression, complementing existing explanations based on data properties or architectural factors.

阅读与讨论 → 访问原文 →

25.

arXiv (CS.LG) 2026-06-18 DOI: arXiv:2606.19057

Quantifying and Auditing LLM Evaluation via Positive–Unlabeled Learning

作者:

Zilong Zhang↗Yi-Ting Hung↗Lei Ding↗Chi-Kuang Yeh↗

arXiv:2606.19057v1 Announce Type: cross Abstract: Large Language Models (LLMs) are increasingly used as judges for scalable evaluation, yet such LLM–as–a–Judge systems exhibit systematic biases that are decoupled from semantic quality, most notably verbosity bias. Meanwhile, human supervision is costly and typically selective, yielding reliable positive judgments but leaving most outputs unlabelled and potentially mixed in quality. We formulate LLM evaluation under selective human supervision as a positive–unlabelled learning problem and propose a geometric auditing framework based on Partial Optimal Transport. By aligning a small set of human–verified positives with a reliable subset of unlabelled outputs in a fixed embedding space, our method identifies human–consistent preferences and corrects biased judges without retraining. Experiments demonstrate improved alignment with human preferences, increased robustness to presentation biases, and interpretable confidence estimates, offering a scalable and statistically grounded alternative to existing LLM–as–a–judge pipelines.

阅读与讨论 → 访问原文 →