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

Recent advances in deep reinforcement learning have made significant stridesin performance on applications such as Go and Atari games. However, developingpractical methods to balance exploration and exploitation in complex domainsremains largely unsolved. Thompson Sampling and its extension to reinforcementlearning provide an elegant approach to exploration that only requires accessto posterior samples of the model. At the same time, advances in approximateBayesian methods have made posterior approximation for flexible neural networkmodels practical. Thus, it is attractive to consider approximate Bayesianneural networks in a Thompson Sampling framework. To understand the impact ofusing an approximate posterior on Thompson Sampling, we benchmarkwell-established and recently developed methods for approximate posteriorsampling combined with Thompson Sampling over a series of contextual banditproblems. We found that many approaches that have been successful in thesupervised learning setting underperformed in the sequential decision-makingscenario. In particular, we highlight the challenge of adapting slowlyconverging uncertainty estimates to the online setting.

Deep Bayesian Bandits Showdown: An Empirical Comparison of Bayesian Deep Networks for Thompson Sampling