AI Paper Atlas

Research map / Reinforcement learning / Offline RL

Tackling Heavy-Tailed Q-Value Bias in Offline-to-Online Reinforcement Learning with Laplace-Robust Modeling

Ruibo Guo, Lei Liu, Rui Yang, Junjie Shen, Guoping Wu, Jie Wang, Bin Li

ICLR 2026 · Reinforcement learning › Offline RL · approach: Offline-to-online fine-tuning

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 LAROO instead of BOORLoffline-to-online RL · D4RL MuJoCoaverage score: +54.8%Improved
use LAROO instead of BOORLoffline-to-online RL · walker2d-randomnormalized return: 6.4 -> 71.6Improved
use LAROO instead of BOORLoffline-to-online RL · D4RL MuJoCo, all tasksdelta_sum(0.1M): 355.4 -> 550.4Improved
use LAROO instead of BOORLoffline-to-online RL · halfcheetah-randomnormalized return: 90.6 -> 87.4Hurt
use LAROO instead of BOORLoffline-to-online RL · hopper-expertnormalized return: 103.2 -> 112.0Improved
replace l2 loss with Laplace-based Db(x) loss in TD3, Cal-QL and ENOTOTD3, Cal-QL, ENOTO · walker2d tasksnormalized returnImproved
add ensemble models to LAROO (ablation)LAROO · walker2d-medium-replayQ bias distribution and returnImproved
add Laplace noise model to LAROO (ablation)LAROO · walker2d-medium-replayQ bias distribution and returnImproved
compare Laplace noise model against ensemble models (ablation)LAROO · walker2d-medium-replayperformanceImproved
use LAROO with UTD=1 and ensemble size 1 instead of Cal-QL (ablation)LAROO · D4RL, 13 experimentsnormalized return: 10 out of 13Improved
use Lb(x) loss instead of Huber and Cauchy losses (ablation)LAROO · D4RLnormalized returnImproved
model Q bias with Laplace instead of Gaussian (ablation)LAROO noise model · walker2d-medium-expert-v2fit to empirical Q biasImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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