Efficient Adversarial Training without Attacking: Worst-Case-Aware Robust Reinforcement Learning.
Yongyuan LiangYanchao SunRuijie ZhengFurong HuangPublished in: CoRR (2022)
Keyphrases
- reinforcement learning
- worst case
- computationally efficient
- average case
- training set
- highly efficient
- supervised learning
- model free
- multi agent
- learning process
- data sets
- hidden markov models
- training phase
- robust estimation
- function approximation
- optimal policy
- lower bound
- face recognition
- machine learning
- data mining
- neural network