A survey paper titled "A Survey of Explainable Reinforcement Learning: Targets, Methods and Needs" reviews over 250 studies in the sub‑domain of eXplainable AI focused on reinforcement learning (XRL). The authors introduce a taxonomy based on the questions "What" (the target of explanation) and "How" (the method of explanation) and use it to organize the state‑of‑the‑art literature. They also highlight related domains that warrant further attention and outline key needs for advancing the field of XRL.
By L\'eo Sauli\`eres
The paper introduces Li‑Net, a new architecture for multi‑channel time series forecasting that captures both linear and non‑linear dependencies among channels. Li‑Net compresses representations across sequence and channel dimensions, processes them with a configurable non‑linear module, and reconstructs forecasts while integrating a sparse Top‑K Softmax attention mechanism within a multi‑scale projection framework. Experiments on real‑world benchmarks show that Li‑Net matches state‑of‑the‑art accuracy while using less memory and achieving faster inference, with ablation studies confirming the contribution of each component.
By Lei Gao, Hengda Bao, Jingfei Fang, Guangzheng Wu, Weihua Zhou, Yun Zhou
EgoP$^{2}$C is a new framework that internalizes a plan‑perceive‑compose workflow within a single vision‑language model for spatial reasoning in egocentric video. It trains specialized Planner and Perception teachers via Dual‑LoRA adapters, then fine‑tunes a mixture‑of‑experts student under their supervision, using a segment‑level multi‑teacher on‑policy distillation to guide evidence‑seeking, stopping, and answer composition. The resulting student models, built on Qwen3‑VL‑8B and SenseNova‑SI‑1.3‑8B, outperform their base models by 9.2 and 7.5 percentage points on three benchmarks, while cutting inference FLOPs by about 66% and speeding up inference 1.29–1.49×.
By Chaoyu Li, Arka Sadhu, Yuliang Li, Hyo Jin Kim, Yuning Chai, Pooyan Fazli
EvalResearchBench (ERB) is a benchmark that tests whether AI agents can autonomously design evaluation tasks for recursive self‑improvement. In ERB, a researcher agent selects or synthesizes tasks, builds graders, and refines them through pilot tests before producing a final evaluator that runs within fixed time and API budgets. Experiments with nine researchers and thirteen candidate models show that the best automated evaluators achieve about 75% concordance with target benchmarks, below the 91% ceiling set by target disagreements, and that human‑designed task samples still outperform many automated approaches.
By Yaolun Zhang, Tianyi Xu, Yujie Zhao, Jishen Zhao, Qingyun Wu, Huazheng Wang
AQUA-Net is a lightweight underwater image enhancement model that fuses spatial, frequency, and illumination information through a residual encoder‑decoder architecture with dual auxiliary branches. The frequency fusion encoder incorporates Fourier‑domain cues to preserve textures and details, while the illumination‑aware decoder applies adaptive exposure correction based on a learned illumination map. Experiments on benchmark datasets and a new high‑resolution Mediterranean Sea video set demonstrate that AQUA-Net achieves state‑of‑the‑art performance with fewer parameters, showing strong generalization and robustness for real‑world underwater imaging.
By Munsif Ali, Najmul Hassan, Lucia Ventura, Davide Di Bari, Simonepietro Canese, Jungpil Shin
The paper introduces SAGE, a framework that adapts 3D geometric foundation models using weak supervision from raw Internet videos. SAGE employs a hierarchical mining pipeline to generate informative training trajectories, sparse geometric anchoring via SfM point clouds, and dense differentiable consistency through 3D Gaussian rendering, while also replaying a small amount of supervised data to avoid catastrophic forgetting. Experiments on two different backbones show that SAGE reduces Chamfer Distance on unseen benchmarks by 20–42% and consistently improves performance, demonstrating that Internet video can serve as a scalable source for geometric learning.
By Zihui Gao, Ke Liu, Donny Y. Chen, Duochao Shi, Guosheng Lin, Jia-Wang Bian, Hao Chen, Chunhua Shen
The paper introduces AOI (Autonomous Operations Intelligence), a multi‑agent framework that treats automated site reliability tasks as a structured trajectory learning problem under security constraints. It combines a trainable diagnostic system using Group Relative Policy Optimization, a read‑write separated execution architecture, and a Failure Trajectory Closed‑Loop Evolver that turns unsuccessful runs into corrective supervision. On the AIOpsLab benchmark, AOI achieves significant performance gains, including a 66.3% best@5 success rate and improved average metrics on unseen fault types.
By Pei Yang, Wanyi Chen, Asuka Yuxi Zheng, Xueqian Li, Xiang Li, Haoqin Tu, Jie Xiao, Yifan Pang, Dongdong Zhang, Fuqiang Li, Alfred Long, Lynn Ai, Eric Yang, Bill Shi, Yifan Chen, Shuhan Sun
The paper introduces an interpretable experiential learning model that uses state history and global feedback to learn a behavioral model represented as a transition graph. Transitions in this graph are annotated with utility values and evidence counts, enabling the model to function in resource-constrained settings. The authors evaluate the approach on the OpenAI Gym Atari Breakout benchmark, showing performance comparable to certain neural network-based methods.
By Anton Kolonin
The paper introduces a process‑aware evaluation framework for cooperative multi‑agent reinforcement learning that supplements traditional outcome metrics with diagnostics of interference, coordinated task completion, participation, and coordination efficiency. Using the STAT testbed and two established benchmarks, the authors demonstrate that policies with similar returns can differ markedly in coordination behavior, and that reward shaping can alter coordination without affecting overall task return. The framework reveals distinct process signatures across spatial, task, and agent scaling and remains informative across different coordination mechanisms.
By Maria Ana Cardei, Matthew Landers, Afsaneh Doryab
STRIDE (Strategic Time-series Reasoning Injected via Distilled Embeddings) is a framework that embeds LLM reasoning directly into the continuous embedding space of Time Series Foundation Models (TSFMs). By distilling reasoning traces into a lightweight LLM and projecting its pooled hidden states as a continuous cross‑modal prior, STRIDE integrates semantic reasoning without fragmenting numerical values or inflating sequence lengths. The method is jointly optimized with cross‑entropy and quantile losses, can be attached to various TSFM architectures, and achieves top performance on the GIFT‑Eval leaderboard and strong results on TFRBench, with ablations showing that the gains come from strategic reasoning integration.
By Md Atik Ahamed, Mihir Parmar, Palash Goyal, Chun-Liang Li, Qiang Cheng, Tomas Pfister, Jinsung Yoon
Lean4AgenticRL is a new framework that introduces multi‑level process rewards grounded in the Lean4 formalization schema to improve agentic reinforcement learning for large language models. It constructs a problem‑specific workflow system that balances capability and controllability, enabling verification of agent behavior and providing rewards at low‑, mid‑, and high‑levels. Experiments show that models trained with Lean4AgenticRL, such as LAR‑27B, outperform comparable models, achieving 77.3% on TerminalBench‑2.1 and surpassing the base model by an average of 9.89%.
By Ruida Wang, Gerald Friedland, Tong Zhang, Chris Kong
arXiv:2610. 22621v1 Announce Type: new Abstract: The training loss of binary and ternary neural networks, expanded as a multilinear polynomial over the discrete weights, is structurally equivalent to the energy of a hierarchical high-order Ising model.
By Zihao Chen, Zhili Xiao, Rui-Jie Zhu, Yipeng Wang, Jason Eshraghian, Shantanu Chakrabartty
The paper introduces Normalizing Flow Maps (NFM), a new class of exactly invertible pushforward maps that extend normalizing flows and flow maps by learning the integrator of an underlying dynamical system while enabling efficient exact likelihood evaluation. NFMs incorporate two consistency axes—pointwise matching for β1-based regression losses and distributional matching for aligning pushforward distributions with a forward process—allowing scalable training. Experiments on CIFAR-10, ImageNet-64, and Boltzmann Generators demonstrate that NFMs achieve state‑of‑the‑art sample quality among parameter‑matched normalizing flows and outperform prior normalizing flows on peptide benchmarks.
By Francesco M. Ruscio, Timothy James Hitge, Fred Zhangzhi Peng, Danyal Rehman, Zander W. Blasingame, Nicholas M. Boffi, Michael M. Bronstein, Alexander Tong, Avishek Joey Bose
arXiv:2610. 22794v1 Announce Type: new Abstract: A prominent family of low-rank LLM optimizers, pioneered by GaLore, performs gradient-subspace projection: at each step, they project the gradient into a low-dimensional subspace and map the update back to the full model.
By Xinyu Luo, Jincheng Zhou, Mengbo Wang, Zilin Shen, Nadia Atallah Lanman, Petros Drineas, Bruno Ribeiro, Brian Bullins
The paper introduces Lookahead Editing (LED), a method that enhances masked diffusion language models (MDMs) by adding a rewritable proposal state. LED performs multiple lookahead updates at each decoding step to refine future predictions before committing tokens, thereby improving expressivity while retaining MDMs’ simple transition schedule. Fine‑tuned LED models (LED‑7B and LED‑8B) achieve significant performance gains on eight benchmarks and outperform larger models such as LLaDA2.1‑16B using only supervised fine‑tuning.
By Yuyan Zhou, Zeyuan Yang, Yuxuan Song, Zhicheng Cai, Hongli Yu, Weiying Ma, Hao Zhou, James Kwok
FLARE is a physics‑grounded benchmark for evaluating language‑model agents on feeder restoration tasks after distribution‑grid faults. It includes 800 AC‑feasible cases across five observation settings, allowing agents to gather evidence, evaluate switches, and perform restoration actions under fixed operational limits. In tests of five instruction‑tuned models over 78,750 episodes, mean normalized regret ranged from 0.082 to 0.148, with a high P95 regret indicating a significant failure tail, and 96.4% of guard‑rejected actions would have caused physical violations.
By BHVSP Subrahmanyam, Neeraj Choudhary
The paper introduces SPLIT, a method that optimally allocates test‑time compute among sampling, model selection, and verification for language models and agents. Using an exact probability formula that accounts for task difficulty, shared blind spots, and verifier correlations, SPLIT is fitted via empirical Bayes on a small labeled pilot and then used to plan budgets for unseen tasks. Across seven diverse benchmarks, 26 models, and six verifier types, SPLIT achieves near‑optimal performance, outperforming rule‑based baselines by up to 2.5 percentage points and providing insights into when additional LLM judges reduce precision.
By Srinivasan Iyengar
The paper presents a deep learning framework, WGAN‑CLP, that forecasts voltage time‑series under anomalies by employing a Wasserstein GAN with a progressive Lipschitz penalty and an adaptive kernel critic. It processes a sliding window of data, predicts subsequent values, reconstructs the stream, and dynamically adjusts its kernel for each new window. Experiments on voltage and energy datasets show the model surpasses benchmark methods in accuracy and robustness, as measured by power ratios, correlation coefficient, and relative RMSE.
By Najmieh Sadat Safarabadi
Residual Stochastic Concept Bottleneck Models (Res‑SCBMs) extend concept‑based models by jointly modeling known concepts and residual variables using a multivariate Gaussian distribution. The learned covariance captures dependencies between concepts and residuals, enabling test‑time concept interventions to influence statistically dependent residuals through Gaussian conditioning. Experiments on concept‑incomplete benchmarks show that Res‑SCBMs outperform competing baselines while preserving effective concept interventions.
By Stephen Marcou, Mateo Espinosa Zarlenga, Sonia Laguna, Moritz Vandenhirtz, Julia E. Vogt, Mateja Jamnik
Blitz is a few‑step distillation framework that accelerates diffusion‑based protein structure prediction models by reducing denoising steps from 200 to eight. It employs a geometric noise schedule, multistep consistency training, physical steering, and target‑free candidate selection to maintain high structural accuracy. Applied to AlphaFold 3 and Boltz‑2, Blitz matches teacher‑level accuracy and outperforms other few‑step baselines across multiple evaluation metrics.
By Vinay Saji Mathew, Soundar Kumara, Gretta Kellogg, William K. M. Lai