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

Research in model-based reinforcement learning has made significant progressin recent years. Compared to single-agent settings, the exponential dimensiongrowth of the joint state-action space in multi-agent systems dramaticallyincreases the complexity of the environment dynamics, which makes it infeasibleto learn an accurate global model and thus necessitates the use of agent-wiselocal models. However, during multi-step model rollouts, the prediction of onelocal model can affect the predictions of other local models in the next step.As a result, local prediction errors can be propagated to other localities andeventually give rise to considerably large global errors. Furthermore, sincethe models are generally used to predict for multiple steps, simply minimizingone-step prediction errors regardless of their long-term effect on other modelsmay further aggravate the propagation of local errors. To this end, we proposeModels as AGents (MAG), a multi-agent model optimization framework thatreversely treats the local models as multi-step decision making agents and thecurrent policies as the dynamics during the model rollout process. In this way,the local models are able to consider the multi-step mutual affect between eachother before making predictions. Theoretically, we show that the objective ofMAG is approximately equivalent to maximizing a lower bound of the trueenvironment return. Experiments on the challenging StarCraft II benchmarkdemonstrate the effectiveness of MAG.

Models as Agents: Optimizing Multi-Step Predictions of Interactive Local Models in Model-Based Multi-Agent Reinforcement Learning