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

In this study, we explore the robustness of cooperative multi-agentreinforcement learning (c-MARL) against Byzantine failures, where any agent canenact arbitrary, worst-case actions due to malfunction or adversarial attack.To address the uncertainty that any agent can be adversarial, we propose aBayesian Adversarial Robust Dec-POMDP (BARDec-POMDP) framework, which viewsByzantine adversaries as nature-dictated types, represented by a separatetransition. This allows agents to learn policies grounded on their posteriorbeliefs about the type of other agents, fostering collaboration with identifiedallies and minimizing vulnerability to adversarial manipulation. We define theoptimal solution to the BARDec-POMDP as an ex post robust Bayesian Markovperfect equilibrium, which we proof to exist and weakly dominates theequilibrium of previous robust MARL approaches. To realize this equilibrium, weput forward a two-timescale actor-critic algorithm with almost sure convergenceunder specific conditions. Experimentation on matrix games, level-basedforaging and StarCraft II indicate that, even under worst-case perturbations,our method successfully acquires intricate micromanagement skills andadaptively aligns with allies, demonstrating resilience against non-obliviousadversaries, random allies, observation-based attacks, and transfer-basedattacks.

Byzantine Robust Cooperative Multi-Agent Reinforcement Learning as a Bayesian Game