AI Paper Atlas

Research map / Reinforcement learning / Offline RL

State Proficiency-Based Adaptive Fine-Tuning for Offline-to-Online Reinforcement Learning

Songlin Li, Wei Xiao, Hao Wu, Xiaodan Zhang, Daolong An, Shuai Lü

AAAI 2026 · Reinforcement learning › Offline RL · approach: Offline-to-online fine-tuning · 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
classify samples by action comparison instead of random classification with matched ratio (ablation)SPA · D4RL Locomotion (8 medium-level tasks)total normalized score: 910.1 -> 961.6Improved
use dual thresholds (offline policy for offline data, historical optimal policy for online data) instead of a single high threshold (ablation)SPA · walker2d-mediumtraining stability, constraint ratio on offline data: 43.01 -> 52.07Improved
use dual thresholds instead of a single low threshold (ablation)SPA · halfcheetah-medium-replayfinal performance, constraint ratio on online data: online constraint ratio 59.27 -> 33.79Improved
add value-function-based classification correction (ablation)SPA · walker2d-medium-replayprecision of high proficiency classification (%): 51.70 -> 60.22Improved
add value-based correction rather than randomly reclassifying the same proportion of samples (ablation)SPA · walker2d-medium-replayprecision (%): 48.15 -> 60.22Improved
relax policy constraint ratio alone (without accurate correction) (ablation)SPA · D4RL LocomotionperformanceNo clear effect
apply SPA to TD3+BC fine-tuning versus baselinesTD3+BC · D4RL Locomotion (12 tasks)total normalized score: 694.4 -> 1142.7Improved
apply SPA to TD3+BC fine-tuning on Maze2DTD3+BC · D4RL Maze2D (6 tasks)total normalized score: 362.4 -> 1115.8Improved
fine-tune with SPA on a task whose offline Q-values explodedSPA · walker2d-random-v2normalized score: 4.2 -> 0.2Hurt
use action-comparison proficiency on halfcheetah-medium compared with random classification (ablation)SPA · halfcheetah-medium-v2normalized score: 60.3 -> 91.7Improved
use action-comparison proficiency on walker2d-medium compared with random classification (ablation)SPA · walker2d-medium-v2normalized score: 116.3 -> 116.1No clear effect

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