BriefGPT.xyz
Jan, 2019
Go-Explore:一种解决困难探索问题的新方法
Go-Explore: a New Approach for Hard-Exploration Problems
HTML
PDF
Adrien Ecoffet, Joost Huizinga, Joel Lehman, Kenneth O. Stanley, Jeff Clune
TL;DR
采用记忆、回归与模拟学习等技术结合的算法Go-Explore在复杂探索问题上有了显著提高,打破了Montezuma's Revenge甚至超越了Pitfall的人类高分纪录,为强化学习领域提供了新思路。
Abstract
A grand challenge in
reinforcement learning
is intelligent
exploration
, especially when rewards are sparse or deceptive. Two Atari games serve as benchmarks for such hard-
→