Policy Smoothing for Provably Robust Reinforcement Learning.
Aounon KumarAlexander LevineSoheil FeiziPublished in: ICLR (2022)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- action selection
- neural network
- worst case
- approximate dynamic programming
- function approximation
- machine learning
- markov decision process
- reinforcement learning algorithms
- state dependent
- learning problems
- smoothing algorithm
- state space
- reinforcement learning problems
- partially observable environments