The paper introduces DiscoveryRSI, a framework that recursively learns to generate valuable scientific hypotheses by building hypothesis trees through multi‑agent evolutionary search, compiling investigations into executable tasks, and refining the process via credit optimization, replay, and rediscovery. It also presents DFM‑Bench, a benchmark for assessing hypothesis quality across multiple dimensions. Experiments in four life‑science domains show that the system improves hypothesis quality, particularly in falsifiability and data resolvability, and outperforms other large‑language‑model agents in expert evaluations.
By Jianyuan Zhong, Chenwei Wu, Zhaochen Yu, Xixian Jiang, Xuewei Li, Hao Xiao, Xilin Dai, Junyang Wang, Jiaru Zou, Zhiqing Cui, Boyuan Zheng, Jing Xiong, Zhen Zhang, Qiang Gao, Michael Morley, Luis Nakayama, Qiang Xu, Liyue Shen, Tao Gui, Leo Anthony Celi, James Zou, Zhenfei Yin, Yingcheng Wu, Ling Yang
SAFE‑Mem is a fine‑grained memory governance framework for large language model agents that annotates each memory claim with governance labels—kind, authority, and scope—to capture provenance and permitted use. It uses governance‑aware retrieval to match queries with claim permissions, blocking unauthorized influence while preserving legitimate information, and builds a governance‑enabled memory graph for provenance tracking and efficient retrieval. Experiments on four LLM backbones show that SAFE‑Mem achieves the lowest attack success rate among defenses that retain legitimate memory and improves answer accuracy by up to 15%.
By Haoming Li, Dingyi Kang, Xu Hu, Jinxin Yang, Zhu Xu, Chenru Wang, Yi Ge, Miao Yin, Guanpeng Li, Feng Chen, Bingzhe Li
Tree-Aware Posterior Distillation for Long-Horizon Agentic Reinforcement Learning proposes TreeOPSD, a critic‑free framework that treats each interaction turn as a tree node, branches alternative continuations, and recursively backs up terminal rewards to generate implicit turn‑level credit. The method maps this credit to a bounded teacher distribution based on the likelihood ratio to a frozen behavior policy, and minimizes the divergence between teacher and student to recover policy‑gradient directions while limiting updates across mini‑batches. Empirical results show TreeOPSD enhances agentic performance on long‑horizon benchmarks, notably improving Qwen3‑30B‑A3B‑Thinking‑2507’s accuracy by 4.97 points over GRPO in the Deep Research setting.
By Heming Zou, Qi Wang, Yixiu Mao, Yun Qu, Yuyi Guo, Zeyuan Liu, Chenglin Li, Weijie Liu, Saiyong Yang, Xiangyang Ji
Computational Multi-Agents Society Experiments (CMASE) is a generative agent‑based framework that enables researchers to participate in‑world and trace interventions at runtime, treating simulation as a simulated ethnographic field. It combines LLM‑driven perception, memory, and emotion with a composable environment and researcher roles that allow real‑time interaction, controlled experiments, and forward exploration of intervention trajectories. Using CMASE, the authors replicated two published field studies—urban greenery and social fragmentation, and a public goods game—demonstrating that embedded social interventions produce observable, auditable processes that parametric manipulation alone does not.
By Hanzhong Zhang, Muhua Huang, Jindong Wang
R$^2$ Flow is a recursive self‑improvement framework for large‑language‑model agents that alternates policy learning, independent verification, and versioned skill‑library updates on a shared‑state orchestration graph. By merging histories that differ only in the order of independent steps, it pools evidence across equivalent executions and uses flow‑share readouts, signed utility ranks, verifier evidence, and residual‑variance plateaus to decide when and which skills to edit. Experiments on question answering, mathematical reasoning, interactive decision making, and code generation show that R$^2$ Flow improves task accuracy and library‑edit precision compared to heuristic orchestration, reinforcement learning, and other skill‑evolution baselines, and it transfers across different executors.
By Mingda Zhang, Qiang Huang, Yanjin Li, Zijia Wang, Qika Lin, Xiaoying Tang, Tiesunlong Shen
The paper introduces CaRDI, a calibrated planning framework for interactive medical diagnosis that uses a clinical loss combining diagnosis error and severity-weighted omission. Unlike prior methods that focus solely on uncertainty reduction, CaRDI balances expected risk reduction with interaction cost, employing diagnostic and safety beliefs along with conformal residual calibration. Experiments on multiple medical benchmarks demonstrate that CaRDI improves diagnostic accuracy, reduces the number of questions asked, and lowers clinical loss by 21.5% while decreasing severe omissions.
By Yue Zhou, Haiyang Zhou, Jin Zhang, Kong Wang, Zihao Rao, Maolin Wang, Hanwen Du, Youhua Li
The paper introduces a personalized memory gating approach for language‑model agents, addressing the challenge that a finite memory bank cannot store every fact. It proposes a retrieval‑based labeling strategy to identify which facts are later used, and formulates online memory gating as a contextual bandit problem with delayed one‑sided feedback. The authors present Warm‑Start UCB, combining offline learning with online adaptation, and provide a theoretical regret bound while experimentally showing superior performance over baselines across various memory budgets and history lengths.
By Kaixuan Ji, Zixiang Chen, Wenting Zhao, Cheng Qian, Heng Wang, Weizhi Zhang, Jielin Qiu, Liangwei Yang, Quanquan Gu, Silvio Savarese, Shelby Heinecke, Huan Wang
The paper introduces STC-MGNet, a spatial‑temporal‑channel attention‑enhanced multimodal graph network designed to predict pedestrian crossing intentions. It employs a pose graph encoder with three attention branches that focus on skeletal joints, observation frames, and feature channels, and integrates visual and ego‑vehicle motion data through a multi‑stage cross‑modal channel modulation mechanism. Experiments on JAADbeh, JAADall, and PIE datasets show accuracy rates of 70%, 87%, and 88%, surpassing several baseline methods.
By Yuanzhe Li, Meng Wang
HEM-Med introduces a hierarchical agentic memory framework for safe multi‑center medication recommendation, addressing two key gaps in existing methods. It disentangles cross‑hospital clinical experience from hospital‑specific prescribing patterns via a hierarchical memory module, and incorporates a safety‑aware medication set generation module that considers drug‑drug interactions and local formulary constraints. Experiments on a real‑world multi‑center eICU dataset demonstrate that HEM‑Med outperforms representative transfer learning, LLM‑based, and MR baselines in both recommendation accuracy and safety.
By Min Xiao, Jin Li, Shoujin Wang, Yeguang Zhen, Yishuo Li, Wenpeng Lu
The paper introduces Heuresis, a benchmark of anonymous simulated datasets spanning linear Gaussian to heavy-tailed mechanisms, designed to evaluate how well LLM agents can perform causal discovery and improve upon existing algorithms. The benchmark tracks algorithmic gains versus data‑driven insights and records every edge an agent modifies. Using this benchmark, GPT‑6.1 Sol and Claude Opus 5.5 outperform the best existing algorithms on all six dataset families, with GPT‑6.1 Sol achieving a 14% adjacency F1 improvement on the hardest family.
By Pingchuan Ma, Rui Ding, Shuai Wang
The paper introduces Grounded Credit Assignment (GCA) for reinforcement learning in multi‑turn search agents. GCA separates credit for evidence retrieval and answer correctness, correcting the bias in GRPO that favors evidence‑retrieving rollouts only proportionally to accuracy gains. Applying GCA to Search‑R1 boosts average EM by 5.7–6.3 points and increases the proportion of evidence‑supported correct answers from 76.9% to 93.4%.
By Qirui Liu, Hao Chen, Yi Wen, Weijie Shi, Yao Zhao, Xianhe Chen, Hao Zhao, Xing Tang, Weihong Luo, Yuyu Luo
The paper investigates how the textual serialization of knowledge graphs influences the performance of large language models (LLMs). It evaluates 15 different graph representation formats across nine advanced LLM configurations on tasks involving ontology-based knowledge extraction and structural graph reasoning. While knowledge extraction shows little sensitivity to serialization, structural reasoning accuracy varies significantly depending on the chosen format, model, query, and task, suggesting that graph serialization should be treated as a dynamic orchestration decision.
By Julien Yu, Mihaela Bornea, Avirup Sil, Katrin Kirchhoff
STRIDE (SpatioTemporal Reasoning In Driving Edge Cases) is a new benchmark designed to evaluate vision-language models on complex, safety-critical driving scenarios. It includes six task families—spatial perception, spatial understanding, temporal memory, temporal extrapolation, trajectory prediction, and scene-context awareness—using multiple-choice questions and free-form visual question answering that require detailed scene explanations and driving decision justifications. The benchmark focuses on unusual object behaviors and complex traffic interactions, and results show that even the best model, GPT‑6‑Astra, scores only 43.4% on average, highlighting significant gaps in spatiotemporal reasoning.
By Lieqi Liu, Rui gao, Jia-Chen Gu, Wenbo Hu, Zhaobin Mo, Ahmadreza Moradipari, Nejib Ammar, Wei Wang, Nanyun Peng
The paper introduces a cost‑aware controller called certified predictive advice gating that decides when to request language‑model advice in reinforcement learning. By forecasting advisor responses, evaluating parsed actions, and reusing cached advice, the gate only requests advice when a calibrated lower bound on the value exceeds the cost. Experiments on BabyAI tasks show that the gate improves return while reducing advice calls by over 97 % compared to always asking, though matched‑budget comparisons do not show a clear advantage over an early‑query schedule.
By Ibne Farabi Shihab, Md Najmus Swaqeeb, Abu Sa-Adat Mohamed Moon-Im Al Ahsan, Joyanta Jyoti Mondal
CastFlow is a role‑specialized agentic forecasting framework that combines general‑purpose tool reasoning with specialized temporal forecasting. It uses an Experience Memory to guide a frozen LLM in acquiring evidence via a multi‑view toolkit, then a lightweight specialized LLM refines predictions through evidence‑grounded reasoning. The two‑stage training strategy—supervised fine‑tuning followed by reinforcement learning with verifiable rewards—optimizes the specialized model based on final forecasting outcomes, achieving superior performance on nine out of ten real‑world benchmarks.
By Bokai Pan, Mingyue Cheng, Xiaoyu Tao, Zhiding Liu, Shuo Yu, Yuchong Wu, Qi Liu, Defu Lian, Enhong Chen
The paper introduces RoboHarness, a simple robotic harness that equips large language models (LLMs) with a visual‑geometric control panel for direct manipulation of embodied tasks. By enabling LLM agents to understand and invoke actions without complex vision‑language‑action (VLA) training, RoboHarness allows models such as Qwen3.8‑Flash‑Next to outperform state‑of‑the‑art VLA models on the BEHAVIOR Challenge 2025 tasks. The authors provide an open‑source implementation of the system.
By Bince Qu, Wei Chen, Bo Zhang
UniField++ is a two‑agent framework that enhances MRI images across multiple field‑strength transitions, such as 64 mT‑to‑3 T and 3 T‑to‑7 T. It combines a VLM‑based planner that predicts risk regions and a flow‑matching enhancer that performs loop reasoning to refine high‑risk areas, guided by radiologist‑annotated hallucination‑prone regions. The method achieves significant improvements in PSNR, SSIM, and whole‑brain segmentation performance compared to existing approaches.
By Yiyang Lin, Chenhui Wang, Tianliang Yao, Shuaibo Li, Zijun Sun, Yiming Deng, Yixuan Yuan
The paper introduces TAG-Nav, a task‑agnostic graph navigation framework that augments mobile GUI agents with reusable destination memory, enabling offline exploration to be distilled into compact records of app functions, target pages, and waypoint routes. It also presents DestBench, a navigation‑only evaluation protocol that isolates route finding from downstream task execution. Experiments show that graph‑derived memory improves navigation success rates for several large‑language‑model agents, demonstrating that destination coverage and route cleanliness are key factors in the value of graph memory.
By Adamo Orsini, Seyed Mehdi Ayyoubzadeh, Sadman Rakib Pinon, Hannuo Zhang, Koosha Gholipour Baradari, Yuanhao Yu, Xinxin Zuo, Yang Wang
arXiv:2610. 23044v1 Announce Type: new Abstract: Rust is increasingly adopted as a memory-safe successor to C, and transpilers such as \crust{} can migrate entire C projects while preserving their behavior.
By Zhiqiang Yuan, Chong Wang, Zhenhao Zhou, Junwei Liu, Xin Peng
Optimist Deep Research is an open‑source agent that conducts literature research as an iterative, state‑centric process. It breaks a research question into a tree of subquestions, uses search intents to query multiple scholarly databases, and merges results into a provenance‑preserving document registry. The agent iteratively expands the registry via citation snowballing, evaluates evidence, and refines the research plan through a critic that can add, split, rephrase, deprioritize, or close subquestions, ultimately producing a report with every citation linked to a retrieved document.
By Gihan Panapitiya