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

The necessity for cooperation among intelligent machines has popularisedcooperative multi-agent reinforcement learning (MARL) in the artificialintelligence (AI) research community. However, many research endeavors havebeen focused on developing practical MARL algorithms whose effectiveness hasbeen studied only empirically, thereby lacking theoretical guarantees. Asrecent studies have revealed, MARL methods often achieve performance that isunstable in terms of reward monotonicity or suboptimal at convergence. Toresolve these issues, in this paper, we introduce a novel framework namedHeterogeneous-Agent Mirror Learning (HAML) that provides a general template forMARL algorithmic designs. We prove that algorithms derived from the HAMLtemplate satisfy the desired properties of the monotonic improvement of thejoint reward and the convergence to Nash equilibrium. We verify thepracticality of HAML by proving that the current state-of-the-art cooperativeMARL algorithms, HATRPO and HAPPO, are in fact HAML instances. Next, as anatural outcome of our theory, we propose HAML extensions of two well-known RLalgorithms, HAA2C (for A2C) and HADDPG (for DDPG), and demonstrate theireffectiveness against strong baselines on StarCraftII and Multi-Agent MuJoCotasks.

Heterogeneous-Agent Mirror Learning: A Continuum of Solutions to Cooperative MARL