arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2608.17947cs.AIcs.LGcs.NE

通过程序搜索与持续抽象发现实现过程式内容元生成

Procedural Content Metageneration via Program Search and Continual Abstraction Discovery

Matthew Siper, Ahmed Khalifa, Julian Togelius

首次发表
浏览论文内容

中文总结 AI 辅助

该研究在推箱子等4款游戏中,提出持续抽象发现(CAD)方法,结合2×2实验,证实其可提升进化程序搜索的适应度,且学习到的库能被后续程序复用。

中文摘要 AI 辅助

大型语言模型能够生成可执行程序,这使得直接在过程式内容生成器而非单个关卡上进行搜索成为可能。我们在推箱子(Sokoban)、塞尔达(Zelda)、危险戴夫(Dangerous Dave)和淘金者(Lode Runner)中研究该方法。每次运行通过语言模型的变异与交叉进化出完整的Python生成器。我们引入持续抽象发现(Continual Abstraction Discovery,简称CAD),该方法从高适应度程序中提取可重用原语,形成特定运行的辅助模块。我们开展2×2实验,将CAD与固定手写领域API的访问权限进行交叉,完成的数据集包含160次完整运行,每个单元至少有10次50代的运行。在所有8个领域与API的比较中,CAD均提升了平均最终最佳适应度;在所有CAD运行中,学习到的库被多数后续程序采用,并反复重新发现验证、可达性及结构效用。这些结果表明,发现可重用原语可改进针对内容生成器的进化程序搜索。

英文摘要

Large language models can generate executable programs, which makes it possible to search directly over procedural content generators rather than individual levels. We study this approach in Sokoban, Zelda, Dangerous Dave, and Lode Runner. Each run evolves complete Python generators through language-model mutation and crossover. We introduce Continual Abstraction Discovery, or CAD, which extracts reusable primitives from high-fitness programs into a run-specific helper module. A 2x2 experiment crosses CAD with access to a fixed hand-written domain API. The completed data set contains 160 complete runs, with at least ten 50-generation runs in every cell. CAD raises mean final best fitness in all eight domain and API comparisons. Across all CAD runs, learned libraries are adopted by most later programs and repeatedly rediscover validation, reachability, and structural utilities. These results support that discovering reusable primitives improves evolutionary program search for content generators.

发表机构

  • New York University(纽约大学)
  • University of Malta(马耳他大学)

机构由 AI 辅助整理,请以论文原文为准。

补充信息

↑