AI Paper Atlas

Research map / Reinforcement learning / Offline RL

Unifying Value Alignment and Assignment in Cross-Domain Offline Reinforcement Learning with Heterogeneous Datasets

Zhongjian Qiao, Jiafei Lyu, Chenjia Bai, Peisong Wang, Siyang Gao, Shuang Qiu

ICML 2026 · Reinforcement learning › Offline RL · approach: Simple practical offline RL · 0 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
add V2AIGDF · D4RL MuJoCo cross-domain (24 tasks, total)total normalized score: 1286.7 -> 1562.5Improved
add V2AOTDF · D4RL MuJoCo cross-domain (24 tasks, total)total normalized score: 1319.5 -> 1612.9Improved
add DVDFIGDF · D4RL MuJoCo cross-domain (24 tasks, total)total normalized score: 1286.7 -> 1374.7Improved
add DVDFOTDF · D4RL MuJoCo cross-domain (24 tasks, total)total normalized score: 1319.5 -> 1395.9Improved
replace DVDF with V2AIGDF · D4RL MuJoCo cross-domain (24 tasks, total)task wins: 20 out of 24 tasksImproved
replace DVDF with V2AOTDF · D4RL MuJoCo cross-domain (24 tasks, total)task wins: 21 out of 24 tasksImproved
vary trade-off coefficient lambda away from 0.6 (0 to 1.0) (ablation)V2A · hopper-me source, hopper-me / hopper-e targetnormalized scoreHurt
vary data selection ratio xi (0.25 to 1.0) (ablation)V2A · halfcheetah-me sourcenormalized scoreMixed
add V2A on halfcheetah-mr -> expertIGDF · halfcheetah-mr source -> medium targetnormalized score: 28.3 -> 26.8Hurt

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