...
  • 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Evolutionary computation >Evolving Multimodal Robot Behavior via Many Stepping Stones with the Combinatorial Multiobjective Evolutionary Algorithm
【24h】

Evolving Multimodal Robot Behavior via Many Stepping Stones with the Combinatorial Multiobjective Evolutionary Algorithm

机译：Evolving Multimodal Robot Behavior via Many Stepping Stones with the Combinatorial Multiobjective Evolutionary Algorithm

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相关主题

摘要

An important challenge in reinforcement learning is to solve multimodal problems,where agents have to act in qualitatively different ways depending on the circumstances.Because multimodal problems are often too difficult to solve directly, it is oftenhelpful to define a curriculum, which is an ordered set of subtasks that can serve asthe stepping stones for solving the overall problem. Unfortunately, choosing an effectiveordering for these subtasks is difficult, and a poor ordering can reduce theperformance of the learning process. Here, we provide a thorough introduction andinvestigation of the Combinatorial Multiobjective Evolutionary Algorithm (CMOEA),which allows all combinations of subtasks to be explored simultaneously.We compareCMOEA against three algorithms that can similarly optimize on multiple subtasks simultaneously:NSGA-II, NSGA-III, and -Lexicase Selection. The algorithms are testedon a function-optimization problem with two subtasks, a simulated multimodal robotlocomotion problem with six subtasks, and a simulated robot maze-navigation problemwhere a hundred random mazes are treated as subtasks. On these problems, CMOEAeither outperforms or is competitive with the controls. As a separate contribution, weshow that adding a linear combination over all objectives can improve the ability of thecontrol algorithms to solve these multimodal problems. Lastly, we show that CMOEAcan leverage auxiliary objectives more effectively than the controls on the multimodallocomotion task. In general, our experiments suggest that CMOEA is a promising algorithmfor solving multimodal problems.

著录项

获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号