Fetching the paper…

Efficient Adversarial Training without Attacking: Worst-Case-Aware Robust Reinforcement Learning · Around