Research map / Reinforcement learning / Visual and representation RL
Reinforcement Learning with Augmented Data
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 |
|---|---|---|---|
| add random data augmentation to pixel SAC | Pixel SAC · DMControl500k Walker walk | score: 42 -> 918 | Helped |
| RAD vs contrastive CURL | SAC · DMControl500k Cheetah run | score: 518 -> 728 | Helped |
| add random crop augmentation | PPO · ProcGen BigFish 100 levels | test return: 1.9 -> 5.4 | Helped |
| add random convolution augmentation (ablation) | PPO · ProcGen StarPilot 100 levels | test return: 18.0 -> 7.4 | Hurt |
| add grayscale augmentation (ablation) | PPO · ProcGen StarPilot 100 levels | test return: 18.0 -> 10.6 | Hurt |
| add cutout-color augmentation (ablation) | PPO · ProcGen StarPilot 100 levels | test return: 18.0 -> 22.4 | Helped |
| add random amplitude scaling to state inputs | SAC · OpenAI Gym Walker | return: -382.5 -> 806.4 | Helped |
| add random amplitude scaling to state inputs | SAC · OpenAI Gym Pendulum | return: 162.1 -> 167.4 | Helped |
| random window (information hiding) augmentation (ablation) | RAD · DMControl Walker walk | score | No clear effect |
| add random translate even with smallest frame (102) (ablation) | SAC · DMControl Cartpole, Cheetah | score | Helped |
Closest papers by meaning
- Understanding when Dynamics-Invariant Data Augmentations Benefit Model-free Reinforcement Learning Updates (ICLR 2024 · similarity 0.90)
- Revisiting Data Augmentation in Deep Reinforcement Learning (ICLR 2024 · similarity 0.90)
- Reinforcement Learning with Euclidean Data Augmentation for State-Based Continuous Control (NeurIPS 2024 · similarity 0.89)
- The Surprising Ineffectiveness of Pre-Trained Visual Representations for Model-Based Reinforcement Learning (NeurIPS 2024 · similarity 0.88)
- Zero Shot Generalization of Vision-Based RL Without Data Augmentation (ICML 2025 · similarity 0.88)
- MAD-TD: Model-Augmented Data stabilizes High Update Ratio RL (ICLR 2025 · similarity 0.88)
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.