Research map / Reinforcement learning / Visual and representation RL
Learning Better with Less: Effective Augmentation for Sample-Efficient Visual Reinforcement Learning
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 |
|---|---|---|---|
| replace PadCrop with Random PadResize as DA | DrQ-V2 · DMC 12-task average | average episode return: 547.96 -> 588.75 | Helped |
| replace PadCrop with Random PadResize as DA | DrQ-V2 · CARLA 200k average | average episode return: 217.8 -> 227.7 | Helped |
| add CycAug cycling between PadCrop and Rand PR | DrQ-V2 · DMC 12-task average | average episode return: 547.96 -> 654.83 | Helped |
| add CycAug cycling between PadCrop and Rand PR | DrQ-V2 · CARLA 100k average | average episode return: 99.7 -> 143.3 | Helped |
| add CycAug cycling between PadCrop and Rand PR | DrQ-V2 · CARLA 200k average | average episode return: 217.8 -> 269.7 | Helped |
| increase DA hardness (ablation) | individual DA operations on DrQ-V2 · DMC tasks | IQM return | Hurt |
| increase spatial diversity of DA (ablation) | individual DA operations on DrQ-V2 · DMC tasks | IQM return | Helped |
| increase strength diversity of DA (ablation) | individual DA operations on DrQ-V2 · DMC tasks | IQM return | Mixed |
| use composing/sampling/mixing fusion of two DA types instead of an individual operation (ablation) | DrQ-V2 · DMC Quadruped Run | episode return | Hurt |
| add a third DA (CS) to CycAug PC+PR (ablation) | CycAug · DMC | sample efficiency | Helped |
Closest papers by meaning
- Revisiting Data Augmentation in Deep Reinforcement Learning (ICLR 2024 · similarity 0.91)
- Understanding when Dynamics-Invariant Data Augmentations Benefit Model-free Reinforcement Learning Updates (ICLR 2024 · similarity 0.90)
- The Surprising Ineffectiveness of Pre-Trained Visual Representations for Model-Based Reinforcement Learning (NeurIPS 2024 · similarity 0.87)
- Revisiting Plasticity in Visual Reinforcement Learning: Data, Modules and Training Stages (ICLR 2024 · similarity 0.87)
- VIGOR: Zero-Shot Visual Generalization via Latent-Space Consistency in Model-Based Reinforcement Learning (NeurIPS 2026 · similarity 0.87)
- Zero Shot Generalization of Vision-Based RL Without Data Augmentation (ICML 2025 · 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.