发帖
雷达卡
MURAL: Meta-Learning Uncertainty-Aware Rewards for Outcome-Driven
Reinforcement Learning
Kevin Li * 1 Abhishek Gupta * 1 Vitchyr Pong 1 Ashwin Reddy 1 Aurick Zhou 1 Justin Yu 1 Sergey Levine 1
Abstract
Exploration in reinforcement learning is, in gen-
eral, a challenging problem. A common technique
to make learning easier is providing demonstra-
tions from a human supervisor, but such demon-
strations can be expensive and time-consuming to
acquire. I ...
京公网安备 11010802022788号
