Counterfactual State Explanations for Reinforcement Learning Agents via Generative Deep Learning

Abstract

Counterfactual explanations, which deal with "why not?" scenarios, can provide insightful explanations to an AI agent's behavior. In this work, we focus on generating counterfactual explanations for deep reinforcement learning (RL) agents which operate in visual input environments like Atari. We introduce counterfactual state explanations, a novel example-based approach to counterfactual explanations based on generative deep learning. Specifically, a counterfactual state illustrates what minimal change is needed to an Atari game image such that the agent chooses a different action. We also evaluate the effectiveness of counterfactual states on human participants who are not machine learning experts. Our first user study investigates if humans can discern if the counterfactual state explanations are produced by the actual game or produced by a generative deep learning approach. Our second user study investigates if counterfactual state explanations can help non-expert participants identify a flawed agent; we compare against a baseline approach based on a nearest neighbor explanation which uses images from the actual game. Our results indicate that counterfactual state explanations have sufficient fidelity to the actual game images to enable non-experts to more effectively identify a flawed RL agent compared to the nearest neighbor baseline and to having no explanation at all.

arXiv commentFull source code available at https://github.com/mattolson93/counterfactual-state-explanations

References in corpus (24)

• Adam: A Method for Stochastic Optimization
• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• Auto-Encoding Variational Bayes
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
• Axiomatic Attribution for Deep Networks
• Striving for Simplicity: The All Convolutional Net
• Learning Important Features Through Propagating Activation Differences
• Understanding Black-box Predictions via Influence Functions
• Explaining Machine Learning Classifiers through Diverse Counterfactual Explanations
• OpenAI Gym
• Sanity Checks for Saliency Maps
• Deep Recurrent Quantization for Generating Sequential Binary Codes
• Beyond Product Quantization: Deep Progressive Quantization for Image Retrieval
• Model-Based Reinforcement Learning for Atari
• Adversarial Autoencoders
• Fader Networks: Manipulating Images by Sliding Attributes
• Visualizing and Understanding Atari Agents
• Interestingness Elements for Explainable Reinforcement Learning: Understanding Agents' Capabilities and Limitations
• Programmatically Interpretable Reinforcement Learning
• Explaining Image Classifiers by Counterfactual Generation
• Enabling Robots to Communicate their Objectives
• Exploratory Not Explanatory: Counterfactual Analysis of Saliency Maps for Deep Reinforcement Learning
• Learning Finite State Representations of Recurrent Policy Networks
• Counterfactual States for Atari Agents via Generative Deep Learning

Cited by in corpus (5)

• Adversarial attacks and defenses in explainable artificial intelligence: A survey
• Multi-Agent Reinforcement Learning: Methods, Applications, Visionary Prospects, and Challenges
• Analysing Deep Reinforcement Learning Agents Trained with Domain Randomisation
• The Intriguing Relation Between Counterfactual Explanations and Adversarial Examples
• Experiential Explanations for Reinforcement Learning