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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Reinforcement Learning (RL) techniques have drawn great attention in manychallenging tasks, but their performance deteriorates dramatically when appliedto real-world problems. Various methods, such as domain randomization, havebeen proposed to deal with such situations by training agents under differentenvironmental setups, and therefore they can be generalized to differentenvironments during deployment. However, they usually do not incorporate theunderlying environmental factor information that the agents interact withproperly and thus can be overly conservative when facing changes in thesurroundings. In this paper, we first formalize the task of adapting tochanging environmental dynamics in RL as a generalization problem usingContextual Markov Decision Processes (CMDPs). We then propose the AsymmetricActor-Critic in Contextual RL (AACC) as an end-to-end actor-critic method todeal with such generalization tasks. We demonstrate the essential improvementsin the performance of AACC over existing baselines experimentally in a range ofsimulated environments.

AACC: Asymmetric Actor-Critic in Contextual Reinforcement Learning