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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Evaluating the worst-case performance of a reinforcement learning (RL) agentunder the strongest/optimal adversarial perturbations on state observations(within some constraints) is crucial for understanding the robustness of RLagents. However, finding the optimal adversary is challenging, in terms of bothwhether we can find the optimal attack and how efficiently we can find it.Existing works on adversarial RL either use heuristics-based methods that maynot find the strongest adversary, or directly train an RL-based adversary bytreating the agent as a part of the environment, which can find the optimaladversary but may become intractable in a large state space. This paperintroduces a novel attacking method to find the optimal attacks throughcollaboration between a designed function named "actor" and an RL-based learnernamed "director". The actor crafts state perturbations for a given policyperturbation direction, and the director learns to propose the best policyperturbation directions. Our proposed algorithm, PA-AD, is theoreticallyoptimal and significantly more efficient than prior RL-based works inenvironments with large state spaces. Empirical results show that our proposedPA-AD universally outperforms state-of-the-art attacking methods in variousAtari and MuJoCo environments. By applying PA-AD to adversarial training, weachieve state-of-the-art empirical robustness in multiple tasks under strongadversaries. The codebase is released athttps://github.com/umd-huang-lab/paad_adv_rl.

Who Is the Strongest Enemy? Towards Optimal and Efficient Evasion Attacks in Deep RL