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

Deep reinforcement-learning methods have achieved remarkable performance onchallenging control tasks. Observations of the resulting behavior give theimpression that the agent has constructed a generalized representation thatsupports insightful action decisions. We re-examine what is meant bygeneralization in RL, and propose several definitions based on an agent'sperformance in on-policy, off-policy, and unreachable states. We propose a setof practical methods for evaluating agents with these definitions ofgeneralization. We demonstrate these techniques on a common benchmark task fordeep RL, and we show that the learned networks make poor decisions for statesthat differ only slightly from on-policy states, even though those states arenot selected adversarially. Taken together, these results call into questionthe extent to which deep Q-networks learn generalized representations, andsuggest that more experimentation and analysis is necessary before claims ofrepresentation learning can be supported.

Measuring and Characterizing Generalization in Deep Reinforcement Learning