Fig. 8
From: Adversarial attack and defense in reinforcement learning-from AI security view
Examples for adversarial examples successfully attack. The examples show that the method proposed in this paper do have ability to find vulnerabilities under VIN pathfinding, and thus interfere the performance of agent automatic pathfinding. a Sample of testing set. b Available Obstacle 1. c Available Obstacle 2. d Available Obstacle 3. e Available Obstacle 4. f Available Obstacle 5