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

Equipped with the trained environmental dynamics, model-based offlinereinforcement learning (RL) algorithms can often successfully learn goodpolicies from fixed-sized datasets, even some datasets with poor quality.Unfortunately, however, it can not be guaranteed that the generated samplesfrom the trained dynamics model are reliable (e.g., some synthetic samples maylie outside of the support region of the static dataset). To address thisissue, we propose Trajectory Truncation with Uncertainty (TATU), whichadaptively truncates the synthetic trajectory if the accumulated uncertaintyalong the trajectory is too large. We theoretically show the performance boundof TATU to justify its benefits. To empirically show the advantages of TATU, wefirst combine it with two classical model-based offline RL algorithms, MOPO andCOMBO. Furthermore, we integrate TATU with several off-the-shelf model-freeoffline RL algorithms, e.g., BCQ. Experimental results on the D4RL benchmarkshow that TATU significantly improves their performance, often by a largemargin. Code is available here.

Uncertainty-driven Trajectory Truncation for Data Augmentation in Offline Reinforcement Learning