AI Paper Atlas

Research map / Bandits and online learning / Contextual bandits and Thompson sampling

Doubly-Robust Lasso Bandit

Gi-Soo Kim, Myunghee Cho Paik

NeurIPS 2019 · Bandits and online learning › Contextual bandits and Thompson sampling · approach: Oracle-efficient contextual bandits · 23 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use doubly-robust pseudo-rewards with average contexts instead of fitting only to the chosen arm's context and rewardDR Lasso Bandit · Simulations with correlated arm contextscumulative regretImproved
Increase correlation among arm contextsDR Lasso Bandit versus Lasso Bandit · Simulations, strong versus weak context correlationcumulative regretImproved
Use DR Lasso Bandit rather than Lasso BanditDR Lasso Bandit · Simulations with 10 armsconvergence to the optimal decision ruleHurt
Use DR Lasso Bandit rather than Lasso BanditDR Lasso Bandit · Simulations across varying numbers of armssensitivity of regret to number of armsImproved
Use inverse-probability weighting instead of the doubly-robust estimatorDR Lasso Bandit pseudo-reward estimation · Stochastic linear contextual banditpseudo-reward varianceHurt
Impose a constant lower bound on arm-selection probabilities to keep inverse-probability-weighted estimates' variance constantIPW-based bandit selection · Stochastic linear contextual banditregretHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy