Munchausen Reinforcement Learning
NIPS 2020, 2020.
We extended the Munchausen idea to distributional Reinforcement Learning, showing that it could be successfully combined with Implicit Quantile Network to outperform the Rainbow baseline
Bootstrapping is a core mechanism in Reinforcement Learning (RL). Most algorithms, based on temporal differences, replace the true value of a transiting state by their current estimate of this value. Yet, another estimate could be leveraged to bootstrap RL: the current policy. Our core contribution stands in a very simple idea: adding t...More
PPT (Upload PPT)