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

Offline reinforcement learning leverages previously-collected offlinedatasets to learn optimal policies with no necessity to access the realenvironment. Such a paradigm is also desirable for multi-agent reinforcementlearning (MARL) tasks, given the increased interactions among agents and withthe enviroment. Yet, in MARL, the paradigm of offline pre-training with onlinefine-tuning has not been studied, nor datasets or benchmarks for offline MARLresearch are available. In this paper, we facilitate the research by providinglarge-scale datasets, and use them to examine the usage of the DecisionTransformer in the context of MARL. We investigate the generalisation of MARLoffline pre-training in the following three aspects: 1) between single agentsand multiple agents, 2) from offline pretraining to the online fine-tuning, and3) to that of multiple downstream tasks with few-shot and zero-shotcapabilities. We start by introducing the first offline MARL dataset withdiverse quality levels based on the StarCraftII environment, and then proposethe novel architecture of multi-agent decision transformer (MADT) for effectiveoffline learning. MADT leverages transformer's modelling ability of sequencemodelling and integrates it seamlessly with both offline and online MARL tasks.A crucial benefit of MADT is that it learns generalisable policies that cantransfer between different types of agents under different task scenarios. OnStarCraft II offline dataset, MADT outperforms the state-of-the-art offline RLbaselines. When applied to online tasks, the pre-trained MADT significantlyimproves sample efficiency, and enjoys strong performance both few-short andzero-shot cases. To our best knowledge, this is the first work that studies anddemonstrates the effectiveness of offline pre-trained models in terms of sampleefficiency and generalisability enhancements in MARL.

Offline Pre-trained Multi-Agent Decision Transformer: One Big Sequence Model Tackles All SMAC Tasks