CopyCAT: Taking Control of Neural Policies with Constant Attacks

Abstract

We propose a new perspective on adversarial attacks against deep reinforcement learning agents. Our main contribution is CopyCAT, a targeted attack able to consistently lure an agent into following an outsider's policy. It is pre-computed, therefore fast inferred, and could thus be usable in a real-time scenario. We show its effectiveness on Atari 2600 games in the novel read-only setting. In this setting, the adversary cannot directly modify the agent's state -- its representation of the environment -- but can only attack the agent's observation -- its perception of the environment. Directly modifying the agent's state would require a write-access to the agent's inner workings and we argue that this assumption is too strong in realistic settings.

AAMAS 2020

References in corpus (7)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm
• Rainbow: Combining Improvements in Deep Reinforcement Learning
• Robust Adversarial Reinforcement Learning
• Defensive Distillation is Not Robust to Adversarial Examples
• Dopamine: A Research Framework for Deep Reinforcement Learning
• Delving into adversarial attacks on deep policies

Cited by in corpus (5)

• When and How to Fool Explainable Models (and Humans) with Adversarial Examples
• Towards Resilient Artificial Intelligence: Survey and Research Issues
• Adversarial Attacks on Linear Contextual Bandits
• Stealthy and Efficient Adversarial Attacks against Deep Reinforcement Learning
• CROP: Certifying Robust Policies for Reinforcement Learning through Functional Smoothing