Research map / Reinforcement learning / Policy gradient and actor-critic
SPAR: Support-Preserving Action Rectification
Jiaxin Zhao, Weihang Pan, Xun Liang, Binbin Lin
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add latent self-imitation to residual policy (ablation) | SPAR residual · Pen-Cloned | normalized score: 67.8 -> 76.2 | Improved |
| add latent self-imitation to residual policy (ablation) | SPAR residual · Hopper medium-replay | normalized score: 72.7 -> 101.9 | Improved |
| add latent self-imitation to residual policy (ablation) | SPAR residual · AntMaze Large-diverse | normalized score: 40.0 -> 71.0 | Improved |
| restrict self-imitation to the anchored residual space rather than global action space (ablation) | SPAR · Hopper medium-replay | normalized score: 12.1 -> 101.9 | Improved |
| restrict self-imitation to the anchored residual space rather than global action space (ablation) | SPAR · AntMaze Large-diverse | normalized score: 0.0 -> 71.0 | Improved |
| add Stage III value-gated rectification (ablation) | SPAR · Hopper medium-replay | normalized score: 49.5 -> 101.9 | Improved |
| add Stage III value-gated rectification (ablation) | SPAR · AntMaze Large-diverse | normalized score: 60.0 -> 71.0 | Improved |
| add Stage III value-gated rectification (ablation) | SPAR · HalfCheetah medium-expert | normalized score: 96.1 -> 97.0 | Improved |
| add uncertainty penalty (lambda_u=0 to 0.5) (ablation) | SPAR-PROJ · Pen-Cloned | normalized score: 39.2 -> 76.2 | Improved |
| add uncertainty penalty (lambda_u=0 to 0.5) (ablation) | SPAR-MLP · Hopper medium-replay | normalized score: 55.2 -> 101.9 | Improved |
| increase uncertainty penalty from 0.5 to 2 (ablation) | SPAR-PROJ · AntMaze Large-diverse | normalized score: 71.0 -> 60.0 | Hurt |
| use soft rather than hard advantage filtering (ablation) | SPAR-PROJ · AntMaze Large-diverse | normalized score: 30.0 -> 71.0 | Improved |
Closest papers by meaning
- DeFlow: Decoupling Behavior-Prior Modeling and Value Maximization for Offline Policy Extraction (NeurIPS 2026 · similarity 0.88)
- Reflective Prompted Policy Optimization: Trajectory-Grounded Revision and Salience Bias (NeurIPS 2026 · similarity 0.87)
- Iteratively Refined Behavior Regularization for Offline Reinforcement Learning (NeurIPS 2024 · similarity 0.87)
- ACE: Off-Policy Actor-Critic with Causality-Aware Entropy Regularization (ICML 2024 · similarity 0.87)
- A Snapshot of Influence: A Local Data Attribution Framework for Online Reinforcement Learning (NeurIPS 2025 · similarity 0.87)
- Adaptive Advantage-Guided Policy Regularization for Offline Reinforcement Learning (ICML 2024 · similarity 0.87)
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.
