Exploratory Not Explanatory: Counterfactual Analysis of Saliency Maps for Deep Reinforcement Learning
arXiv:1912.05743
Abstract
Saliency maps are frequently used to support explanations of the behavior of deep reinforcement learning (RL) agents. However, a review of how saliency maps are used in practice indicates that the derived explanations are often unfalsifiable and can be highly subjective. We introduce an empirical approach grounded in counterfactual reasoning to test the hypotheses generated from saliency maps and assess the degree to which they correspond to the semantics of RL environments. We use Atari games, a common benchmark for deep RL, to evaluate three types of saliency maps. Our results show the extent to which existing claims about Atari games can be evaluated and suggest that saliency maps are best viewed as an exploratory tool rather than an explanatory tool.
Published at ICLR 2020
References in corpus (7)
- SmoothGrad: removing noise by adding noise
- Attention is not Explanation
- The (Un)reliability of saliency methods
- Counterfactual States for Atari Agents via Generative Deep Learning
- Finding and Visualizing Weaknesses of Deep Reinforcement Learning Agents
- Toybox: A Suite of Environments for Experimental Evaluation of Deep Reinforcement Learning
- ToyBox: Better Atari Environments for Testing Reinforcement Learning Agents