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/

Zendy Plus

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

Zendy Tools

Zendy Open

Zero-shot coordination (ZSC) is a new cooperative multi-agent reinforcementlearning (MARL) challenge that aims to train an ego agent to work with diverse,unseen partners during deployment. The significant difference between thedeployment-time partners' distribution and the training partners' distributiondetermined by the training algorithm makes ZSC a unique out-of-distribution(OOD) generalization challenge. The potential distribution gap betweenevaluation and deployment-time partners leads to inadequate evaluation, whichis exacerbated by the lack of appropriate evaluation metrics. In this paper, wepresent ZSC-Eval, the first evaluation toolkit and benchmark for ZSCalgorithms. ZSC-Eval consists of: 1) Generation of evaluation partnercandidates through behavior-preferring rewards to approximate deployment-timepartners' distribution; 2) Selection of evaluation partners by Best-ResponseDiversity (BR-Div); 3) Measurement of generalization performance with variousevaluation partners via the Best-Response Proximity (BR-Prox) metric. We useZSC-Eval to benchmark ZSC algorithms in Overcooked and Google Research Footballenvironments and get novel empirical findings. We also conduct a humanexperiment of current ZSC algorithms to verify the ZSC-Eval's consistency withhuman evaluation. ZSC-Eval is now available athttps://github.com/sjtu-marl/ZSC-Eval.

ZSC-Eval: An Evaluation Toolkit and Benchmark for Multi-agent Zero-shot Coordination