Snooping Attacks on Deep Reinforcement Learning

Abstract

Adversarial attacks have exposed a significant security vulnerability in state-of-the-art machine learning models. Among these models include deep reinforcement learning agents. The existing methods for attacking reinforcement learning agents assume the adversary either has access to the target agent's learned parameters or the environment that the agent interacts with. In this work, we propose a new class of threat models, called snooping threat models, that are unique to reinforcement learning. In these snooping threat models, the adversary does not have the ability to interact with the target agent's environment, and can only eavesdrop on the action and reward signals being exchanged between agent and environment. We show that adversaries operating in these highly constrained threat models can still launch devastating attacks against the target agent by training proxy models on related tasks and leveraging the transferability of adversarial examples.

arXiv comment13 pages, 12 figures

References in corpus (10)

• Continuous control with deep reinforcement learning
• Playing Atari with Deep Reinforcement Learning
• End to End Learning for Self-Driving Cars
• Trust Region Policy Optimization
• Transferability in Machine Learning: from Phenomena to Black-Box Attacks using Adversarial Samples
• Delving into Transferable Adversarial Examples and Black-box Attacks
• The Space of Transferable Adversarial Examples
• Visualizing and Understanding Atari Agents
• Tactics of Adversarial Attack on Deep Reinforcement Learning Agents
• A Deep Learning Approach for Joint Video Frame and Reward Prediction in Atari Games