Research map / Reinforcement learning / Visual and representation RL
TSTM: Temporal Segmentation for Task-relevant Mask in Visual Reinforcement Learning Generalization
Weicheng Du, Wenjia Meng, Zhengzhe Zhang, Yilong Yin, Xiankai Lu
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 temporal segmentation network (ablation) | TSTM · DMC-GB video hard, Walker Walk | episode return: 391 -> 821 | Improved |
| add temporal segmentation network (ablation) | TSTM · DMC-GB video hard, Cartpole Swingup | episode return: 333 -> 741 | Improved |
| add temporal segmentation network (ablation) | TSTM · DMC-GB video easy, Walker Walk | episode return: 788 -> 912 | Improved |
| add invariant representation loss (ablation) | TSTM · DMC-GB video easy, Walker Walk | episode return: 859 -> 912 | Improved |
| add invariant representation loss (ablation) | TSTM · DMC-GB video hard, Walker Walk | episode return: 770 -> 821 | Improved |
| add policy consistency constraint (ablation) | TSTM · DMC-GB video easy, Walker Walk | episode return: 908 -> 912 | No clear effect |
| add policy consistency constraint (ablation) | TSTM · DMC-GB video hard, Cartpole Swingup | episode return: 469 -> 741 | Improved |
| add policy consistency constraint (ablation) | TSTM · DMC-GB video hard, Walker Walk | episode return: 777 -> 821 | Improved |
| use temporal segmentation instead of single-frame masks (SGQN, MaDi) | TSTM · DMC-GB video hard average | episode return: 688 -> 859 vs SGQN | Improved |
| temporal masks on Walker Stand where the agent barely moves | TSTM · DMC-GB Walker Stand | episode return: 932 -> 923 vs SimGRL | No clear effect |
Closest papers by meaning
- Efficient Reinforcement Learning Through Adaptively Pretrained Visual Encoder (AAAI 2025 · similarity 0.88)
- TTRV: Test-Time Reinforcement Learning for Vision Language Models (CVPR 2026 · similarity 0.87)
- DMC-VB: A Benchmark for Representation Learning for Control with Visual Distractors (NeurIPS 2024 · similarity 0.87)
- Focus-Then-Decide: Segmentation-Assisted Reinforcement Learning (AAAI 2024 · similarity 0.87)
- Learning Latent Dynamic Robust Representations for World Models (ICML 2024 · similarity 0.87)
- Sliding Puzzles Gym: A Scalable Benchmark for State Representation in Visual Reinforcement Learning (ICML 2025 · similarity 0.86)
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.
