AI Paper Atlas

Research map / Reinforcement learning / Offline RL

Less Is More: Clustered Cross-Covariance Control for Offline RL

Nan Qiao, Sheng Yue, Shuning Wang, Yongheng Deng, Ju Ren

ICLR 2026 · Reinforcement learning › Offline RL · approach: Offline RL sample complexity theory · 2 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 C4 instead of best baseline TSRLTSRL · MuJoCo locomotion (10k samples)normalized score: 57.2 -> 75.7Improved
use C4 instead of best baseline BPPOBPPO · Adroit (10k samples)normalized score: 23.1 -> 21.6Hurt
use C4 instead of TSRLTSRL · AntMaze (10k samples)normalized score: 22.0 -> 27.0Improved
use C4 instead of BPPOBPPO · Maze2D (10k samples)normalized score: 120.6 -> 126.9Improved
use C4 instead of TSRL on Halfcheetah-expertTSRL · Halfcheetah-expert (10k)normalized score: 42.0 -> 75.8Improved
use C4 instead of BPPO on Walker2d-medium-expertBPPO · Walker2d-medium-expert (10k)normalized score: 61.3 -> 96.3Improved
use C4 instead of TSRL on Hopper-medium-expertTSRL · Hopper-medium-expert (10k)normalized score: 56.6 -> 81.3Improved
add C4 regularizer to CQL or TD3+BC versus LN and DR3CQL and TD3+BC · MuJoCo locomotion (10k samples)normalized scoreImproved
vary regularization strength lambda and base coefficient alpha (ablation)C4 · MuJoCo locomotionnormalized scoreNo clear effect
vary number of initial clusters and data quality/coverage (ablation)C4 · MuJoCo locomotionnormalized scoreNo clear effect

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