Benchmarking Perturbation-based Saliency Maps for Explaining Deep Reinforcement Learning Agents.
Tobias HuberBenedikt LimmerElisabeth AndréPublished in: CoRR (2021)
Keyphrases
- saliency map
- reinforcement learning agents
- visual attention
- dynamic environments
- visual saliency
- visual saliency detection
- reinforcement learning
- human visual system
- input image
- natural images
- salient object detection
- state abstraction
- salient regions
- video frames
- saliency detection
- visual features
- transfer learning
- multi agent environments
- salient region detection
- low level features
- visual quality
- learning algorithm
- eye movements
- image quality
- vision system
- state space
- active learning
- multi agent systems
- high quality