AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

Policy Likelihood-based Query Sampling and Critic-Exploited Reset for Efficient Preference-based Reinforcement Learning

Jongkook Heo, Jaehoon Kim, Young Jae Lee, Min Gu Kwak, Youngjoon Park, Seoung Bum Kim

ICLR 2026 · Reinforcement learning › RL from feedback · approach: Preference-based RL reward learning

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
reset reward estimator at every feedback sessionPEBBLE · Walker Walk (100/10)episode return and reward overestimation: plateau at approximately 150% without resetImproved
add policy likelihood-based query sampling (PLS) (ablation)PEBBLE · Walker Walk, Door Open, Drawer Openepisode return / success rateImproved
add critic-exploited reset (CER) (ablation)PEBBLE · Walker Walk, Door Open, Drawer Openepisode return / success rateImproved
combine PLS and CER (ablation)PEBBLE · Drawer Open (5000/50)success rate: approximately 100% (CER alone)Hurt
add temporal data augmentation (TA) (ablation)PoLiCER · Door Open, Drawer OpenscoreMixed
add CER on top of PLS (ablation)PLS-only PoLiCER · critic outputcritic overestimationImproved
replace recency-based sampling (QPA) with PLSQPA · Meta-Worldsuccess rate: over 80%Improved
replace QPA with PoLiCER in pixel-based controlQPA · Cheetah Run and Window Open (pixel)performanceImproved
replace PEBBLE with PoLiCERPEBBLE · Window Open (pixel)success rate: approximately 80%, doubling PEBBLEImproved
replace PEBBLE with PoLiCER with human feedbackPEBBLE · Walker Walk and Window Open (human labels)ground truth reward / success rateImproved

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