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

In real-world scenarios, the application of reinforcement learning issignificantly challenged by complex non-stationarity. Most existing methodsattempt to model changes in the environment explicitly, often requiringimpractical prior knowledge. In this paper, we propose a new perspective,positing that non-stationarity can propagate and accumulate through complexcausal relationships during state transitions, thereby compounding itssophistication and affecting policy learning. We believe that this challengecan be more effectively addressed by tracing the causal origin ofnon-stationarity. To this end, we introduce the Causal-Origin REPresentation(COREP) algorithm. COREP primarily employs a guided updating mechanism to learna stable graph representation for states termed as causal-originrepresentation. By leveraging this representation, the learned policy exhibitsimpressive resilience to non-stationarity. We supplement our approach with atheoretical analysis grounded in the causal interpretation for non-stationaryreinforcement learning, advocating for the validity of the causal-originrepresentation. Experimental results further demonstrate the superiorperformance of COREP over existing methods in tackling non-stationarity.

Tackling Non-Stationarity in Reinforcement Learning via Causal-Origin Representation