Research map / Reinforcement learning / Visual and representation RL
Tree-Structured Policy Based Progressive Reinforcement Learning for Temporally Language Grounding in Video
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use a tree-structured policy instead of a flat policy (ablation) | TSP-PRL · Charades-STA | IoU@0.7: 17.13 -> 22.85 | Helped |
| Use a tree-structured policy instead of a flat policy (ablation) | TSP-PRL · ActivityNet | IoU@0.3: 49.35 -> 55.17 | Helped |
| Add the intrinsic reward term to the root reward (ablation) | TSP-PRL · Charades-STA | IoU@0.5: 40.64 -> 45.30 | Helped |
| Add the extrinsic reward term to the root reward (ablation) | TSP-PRL · Charades-STA | IoU@0.5: 44.41 -> 45.30 | Helped |
| Add the alignment network for the stop signal instead of using stop as a root-policy action (ablation) | TSP-PRL · Charades-STA | IoU@0.7: 19.03 -> 24.73 | Helped |
| Increase the maximum testing episode length from 10 to 20 (ablation) | TSP-PRL · Charades-STA | IoU@0.7: 22.85 -> 24.73 | Helped |
| Increase the maximum testing episode length from 20 to 30 (ablation) | TSP-PRL · Charades-STA | IoU@0.7: 24.73 -> 24.75 | No clear effect |
| Use TSP-PRL with C3D features rather than ABLR | Charades-STA · Charades-STA | IoU@0.7: 15.80 -> 17.69 | Helped |
| Use TSP-PRL rather than ABLR | ActivityNet · ActivityNet | MIoU: 36.99 -> 39.21 | Helped |
| Use TSP-PRL with Two-Stream features rather than MAN with I3D features | Charades-STA · Charades-STA | IoU@0.7: 22.72 -> 24.73 | Helped |
Closest papers by meaning
- Read, Watch, and Move: Reinforcement Learning for Temporally Grounding Natural Language Descriptions in Videos (AAAI 2019 · similarity 0.89)
- RLZero: Direct Policy Inference from Language Without In-Domain Supervision (NeurIPS 2025 · similarity 0.87)
- TempSamp-R1: Effective Temporal Sampling with Reinforcement Fine-Tuning for Video LLMs (NeurIPS 2025 · similarity 0.87)
- Text-Aware Diffusion for Policy Learning (NeurIPS 2024 · similarity 0.86)
- Uncertainty-quantified Rollout Policy Adaptation for Unlabelled Cross-domain Video Temporal Grounding (NeurIPS 2025 · similarity 0.86)
- Not Only Where, But When: Temporal Scheduling for RLVR (NeurIPS 2026 · 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.