Mitigation of Policy Manipulation Attacks on Deep Q-Networks with Parameter-Space Noise
This addresses security issues in deep reinforcement learning systems, but it is incremental as it builds on existing noise-based defense methods.
The paper tackles the vulnerability of deep reinforcement learning to policy manipulation attacks via adversarial examples by proposing a mitigation technique based on adding noise to the parameter space during training. The result shows that this technique reduces the transferability of adversarial examples and demonstrates promising performance in mitigating whitebox and blackbox attacks at test and training times.
Recent developments have established the vulnerability of deep reinforcement learning to policy manipulation attacks via intentionally perturbed inputs, known as adversarial examples. In this work, we propose a technique for mitigation of such attacks based on addition of noise to the parameter space of deep reinforcement learners during training. We experimentally verify the effect of parameter-space noise in reducing the transferability of adversarial examples, and demonstrate the promising performance of this technique in mitigating the impact of whitebox and blackbox attacks at both test and training times.