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

Estimating the value function for a fixed policy is a fundamental problem inreinforcement learning. Policy evaluation algorithms---to estimate valuefunctions---continue to be developed, to improve convergence rates, improvestability and handle variability, particularly for off-policy learning. Tounderstand the properties of these algorithms, the experimenter needshigh-confidence estimates of the accuracy of the learned value functions. Forenvironments with small, finite state-spaces, like chains, the true valuefunction can be easily computed, to compute accuracy. For large, or continuousstate-spaces, however, this is no longer feasible. In this paper, we addressthe largely open problem of how to obtain these high-confidence estimates, forgeneral state-spaces. We provide a high-confidence bound on an empiricalestimate of the value error to the true value error. We use this bound todesign an offline sampling algorithm, which stores the required quantities torepeatedly compute value error estimates for any learned value function. Weprovide experiments investigating the number of samples required by thisoffline algorithm in simple benchmark reinforcement learning domains, andhighlight that there are still many open questions to be solved for thisimportant problem.

High-confidence error estimates for learned value functions