arXiv：cs.CL· Amirhossein Abaskohi, Mahdi Mostajabdaveh, Zirui Zhou·· 4 小时前AI 评分33
SEDIMA：面向进化搜索智能体的跨运行分层洞察记忆
SEDIMA: Cross-Run Hierarchical Insight Memory for Evolutionary Search Agents
AI 导读
SEDIMA 是一种面向 LLM 进化搜索智能体的持久化分层洞察记忆，将原始轨迹蒸馏为自然语言洞察，用注意力加权质心按语义相似度聚类，并检索相关指导来约束后续变异，从而跨运行、跨问题积累可迁移知识。
正文
Abstract:Large language model (LLM)-driven evolutionary search is a powerful paradigm for automated program and algorithm discovery, yet existing systems are largely memoryless: each run explores from scratch, so agents repeatedly rediscover the same improvements and re-encounter the same dead ends. We introduce SEDIMA, a persistent hierarchical insight memory for evolutionary search agents. SEDIMA distills raw traces into natural-language insights, clusters them by semantic similarity using attention-weighted centroids, and retrieves relevant guidance to condition future mutations, accumulating transferable knowledge across runs and problems rather than within a single trajectory. As a drop-in module that leaves the search operators unmodified, SEDIMA improves average final performance by 5.5% on AlgoTune and 6.6% on ALE-Bench LITE under a fixed budget of 100 evaluated candidates. Under OpenEvolve, SEDIMA requires 32.3% fewer iterations on average to reach baseline-best performance across the five evaluated backbones.
| Subjects: | Neural and Evolutionary Computing (cs.NE); Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.02361 [cs.NE] |
| (or arXiv:2610.02361v1 [cs.NE] for this version) | |
| https://doi.org/10.48550/arXiv.2610.02361 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Amirhossein Abaskohi [view email]
[v1]
Thu, 1 Oct 2026 18:38:31 UTC (1,072 KB)
来源：arXiv：cs.CL · arxiv.org