As the time horizons of sequential decision-making tasks grow, keeping full interaction histories in model context becomes increasingly costly. Recent work reduces context lengths by instead conditioning decision-making agents on recursively updated natural-language summaries, which are concise and interpretable. However, they underperform agents with access to the full context, suggesting that they fail to generate sufficient summaries. To address this we propose ABBEL, a recursive summarization framework that isolates and directly supervises each summary's information contents in the form of explicit natural-language belief states. First, we analyze the belief states generated by frontier models under ABBEL across five domains, and verify that performance is often degraded due to omitting or incorrectly updating information. We also discover settings where models use memory inefficiently by retaining extraneous information. We target these limitations by fine-tuning with two RL-based methods: belief grading, which reduces update errors by rewarding belief generations based on their information content, and peak belief penalties, which encourage compressing the beliefs with the greatest memory footprints. We demonstrate that these methods significantly reduce the performance gap with full context models, and enable ABBEL to outperform prior memory agent work by 40% while using 67% of the memory. Our code is available at https://github.com/jakob-bjorner/optimal-explorer-dev
Beyond the Shadows of Plato's Cave: Evaluating False Memory in Autonomous Agents via Counterfactual Reasoning
Sydney AI Centre, The University of Sydney(悉尼大学悉尼人工智能中心);Australian Institute for Machine Learning, Adelaide University(阿德莱德大学澳大利亚机器学习研究所);Australian Artificial Intelligence Institute, University of Technology Sydney(悉尼科技大学澳大利亚人工智能研究所);Wuhan University(武汉大学);School of Mathematics and Statistics, The University of Melbourne(墨尔本大学数学与统计学院)
REBA: A Revealed Belief Automaton Framework for Online Planning in Continuous POMDPs
School of Computer Science and Technology, Dalian University of Technology(大连理工大学计算机科学与技术学院);School of Computer Science and Artificial Intelligence, Liaoning Normal University(辽宁师范大学计算机科学与人工智能学院);Human-Centered AI Lab, Department of Ecosystem Management, Climate and Biodiversity, BOKU University(BOKU大学生态系统管理与气候生物多样性系人本人工智能实验室);Institute of Human-Centered Computing, Graz University of Technology(格拉茨技术大学人本计算研究所)
Robust Belief-State Policy Learning for Quantum Network Routing Under Decoherence and Time-Varying Conditions
Department of Electrical Engineering, Columbia University(哥伦比亚大学电气工程系);Department of Electrical Engineering, Imam Khomeini International University(伊玛目·霍梅尼国际大学电气工程系);Department of Electrical Engineering, Qatar University(卡塔尔大学电气工程系)
PRIME: Large Language Model Personalization with Cognitive Dual-Memory and Personalized Thought Process
Computer Science and Engineering, University of Michigan, Ann Arbor, MI(计算机科学与工程系,密歇根大学,安娜堡分校);Department of Political Science, Northeastern University, Boston, MA(政治学系,东北大学,波士顿分校)
该研究提出融合认知双记忆与个性化慢思考的 PRIME 框架,并构建基于 Reddit CMV 的长上下文个性化数据集,实验证明其可捕捉动态用户偏好。
Shared Worlds, Private Minds: Structured Memory for Long-Form Writing as World Creation
Southeast University(东南大学);Beijing Zhongguancun Academy(北京中关村学院);Duke University(杜克大学);East China Normal University(华东师范大学);Renmin University of China(中国人民大学);Communication University of China(中国传媒大学);Jiangsu Second Normal University(江苏第二师范学院);ZhuiWen Technology Co., Ltd.(追闻科技有限公司)
Complexity Reduction in the Negotiation of New Lexical Conventions
University of Bordeaux(波尔多大学);INRIA Bordeaux Sud-Ouest(法国国家信息与自动化研究所波尔多西南中心);Ensta ParisTech(巴黎高等科技学院);SONY Computer Science Lab(索尼计算机科学实验室);Sapienza University of Rome(罗马第一大学);Complexity Science Hub Vienna (CSHV)(维也纳复杂科学中心)