KEEL：从专长与教训中演化知识的持续智能体学习框架
Holding a Steady Course: Knowledge Evolution from Expertise and Lessons for Continual Agent Learning
AI 导读
arXiv 论文提出 KEEL，一个不依赖特定智能体的持续学习框架，利用验证器证据把成功行为转化为可复用技能、把失败转化为纠正性教训，并随新证据累积对技能进行支持、反驳、替换或退役。在 SkillFlow 基准上，KEEL 相比无知识基线将任务成功率提升 4.2 至 21.7 个百分点，高于其默认方法的 2.4 至 9.6 个百分点；消融实验显示技能与教训互补。
正文 · 原文
Abstract:Autonomous agents are increasingly deployed across sequences of related tasks, yet they often fail to retain and improve their knowledge over time. Existing approaches convert interaction trajectories into memories or reusable skills, but experience is not automatically reliable knowledge. Automatically collected skills and memories can propagate errors, mislead, and corrupt useful knowledge. To address this gap, we introduce Knowledge Evolution from Expertise and Lessons (KEEL), an agent-agnostic framework that uses verifier evidence to transform successful behavior into reusable skills and failures into corrective lessons. As new evidence accumulates, individual skill claims can be supported, contradicted, replaced, or retired. On SkillFlow benchmark, KEEL improves task success by 4.2 to 21.7 percentage points over the no-knowledge baseline, compared with gains of 2.4 to 9.6 points from SkillFlow's default method. Ablations show that skills and lessons are complementary. These results show that reliable continual agent learning depends not simply on accumulating experience, but on continually validating and evolving the knowledge derived from it. Our code is available at: this https URL.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.34043 [cs.LG] |
| (or arXiv:2610.34043v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.34043 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Wenzhe Fan [view email]
[v1]
Sun, 25 Oct 2026 04:11:33 UTC (5,751 KB)
来源：arXiv cs.AI · arxiv.org