English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Deep reinforcement learning has shown promising results in learning controlpolicies for complex sequential decision-making tasks. However, these neuralnetwork-based policies are known to be vulnerable to adversarial examples. Thisvulnerability poses a potentially serious threat to safety-critical systemssuch as autonomous vehicles. In this paper, we propose a defense mechanism todefend reinforcement learning agents from adversarial attacks by leveraging anaction-conditioned frame prediction module. Our core idea is that theadversarial examples targeting at a neural network-based policy are noteffective for the frame prediction model. By comparing the action distributionproduced by a policy from processing the current observed frame to the actiondistribution produced by the same policy from processing the predicted framefrom the action-conditioned frame prediction module, we can detect the presenceof adversarial examples. Beyond detecting the presence of adversarial examples,our method allows the agent to continue performing the task using the predictedframe when the agent is under attack. We evaluate the performance of ouralgorithm using five games in Atari 2600. Our results demonstrate that theproposed defense mechanism achieves favorable performance against baselinealgorithms in detecting adversarial examples and in earning rewards when theagents are under attack.

Detecting Adversarial Attacks on Neural Network Policies with Visual Foresight