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

The predictive information, the mutual information between the past andfuture, has been shown to be a useful representation learning auxiliary lossfor training reinforcement learning agents, as the ability to model what willhappen next is critical to success on many control tasks. While existingstudies are largely restricted to training specialist agents on single-tasksettings in simulation, in this work, we study modeling the predictiveinformation for robotic agents and its importance for general-purpose agentsthat are trained to master a large repertoire of diverse skills from largeamounts of data. Specifically, we introduce Predictive Information QT-Opt(PI-QT-Opt), a QT-Opt agent augmented with an auxiliary loss that learnsrepresentations of the predictive information to solve up to 297 vision-basedrobot manipulation tasks in simulation and the real world with a single set ofparameters. We demonstrate that modeling the predictive informationsignificantly improves success rates on the training tasks and leads to betterzero-shot transfer to unseen novel tasks. Finally, we evaluate PI-QT-Opt onreal robots, achieving substantial and consistent improvement over QT-Opt inmultiple experimental settings of varying environments, skills, and multi-taskconfigurations.

PI-QT-Opt: Predictive Information Improves Multi-Task Robotic Reinforcement Learning at Scale