AI Paper Atlas

Research map / Reinforcement learning / Offline RL

Offline Reinforcement Learning with Adaptive Feature Fusion

Tieru Wang, Kunbao Wu, Guoshun Nan

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
set context length to 1 (single-step DT) in toy task (ablation)DT · toy trajectory-stitching tasksuccess rate (%): 0.0 -> 100.0Improved
add Q-value guidance with eta = 0.01 to DT-style model (K=3) (ablation)QT · toy tasksuccess rate (%): 0.1No clear effect
increase QT eta to 10.0 in toy task (ablation)QT · toy tasksuccess rate (%): 0.1 -> 96.2Improved
learn adaptive fusion coefficient instead of fixed alpha-hat = 0 (ablation)QDFFDT · Maze2D+AntMazeaverage score: 86.9 -> 105.3Improved
learn adaptive fusion instead of fixed alpha-hat = 1 (sequence only) (ablation)QDFFDT · Maze2D+AntMazeaverage score: 101.2 -> 105.3Improved
learn fusion instead of fixed alpha-hat = 0.25 (ablation)QDFFDT · Maze2D+AntMazeaverage score: 80.4 -> 105.3Improved
add Q-augmented dual-feature fusion to QTQDFFDT vs QT · D4RL Gym averagenormalized score: 89.8 -> 94.3Improved
add adaptive fusion over QT on AntMazeQDFFDT vs QT · D4RL AntMaze averagenormalized score: 38.9 -> 88.3Improved
use concatenation fusion instead of addition (hopper-medium-replay) (ablation)QDFFDT · hopper-medium-replay-v2normalized score: 102.7 -> 103.9No clear effect
use concatenation fusion instead of addition (walker2d-medium-replay) (ablation)QDFFDT · walker2d-medium-replay-v2normalized score: 101.0 -> 100.3No clear effect
use n-step Bellman instead of 1-step with K=20 (ablation)QDFFDT · halfcheetah-medium-v2normalized score: 65.7 -> 55.6Hurt
use n-step Bellman instead of 1-step with K=20 (ablation)QDFFDT · halfcheetah-medium-replay-v2normalized score: 62.0 -> 38.4Hurt

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