Research map / Reinforcement learning / Visual and representation RL
Predictive Information Accelerates Learning in RL
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 predictive-information CEB auxiliary loss to SAC from pixels (ablation) | SAC · DMC Cartpole Swingup / Cheetah Run (no augmentation) | return | Helped |
| add predictive-information auxiliary loss without data augmentation (ablation) | SAC · DMC Reacher Easy | return | No clear effect |
| add predictive-information auxiliary loss with data augmentation (ablation) | SAC with augmentation · DMC benchmark | return | Helped |
| add random image shift augmentation to CEB learning (ablation) | PI-SAC · DMC Reacher Easy | CEB representation quality / return | Helped |
| add compression (beta>0, e.g. 0.01) to CEB objective (ablation) | PI-SAC · DMC Cartpole Swingup (no augmentation) | return | Helped |
| increase compression to very high beta (ablation) | PI-SAC · DMC tasks | return | Hurt |
| predict the future (PI-SAC) instead of random shift of past (multiview self-prediction) (ablation) | CEB auxiliary task · DMC tasks | return | Helped |
| add compression to MVSP agents (ablation) | MVSP · Cheetah and Reacher | return | Helped |
| use compressed representations when transferring to harder target tasks (ablation) | PI-SAC frozen representation · Cartpole Balance to Swingup, Walker Stand to Walk | return | Helped |
| use compressed representations when transferring to easier target task (ablation) | PI-SAC frozen representation · Walker Walk to Stand | return | No clear effect |
| use CEB representation directly as actor/critic state (Representation PI-SAC) instead of auxiliary loss (ablation) | PI-SAC · Cheetah, Hopper, Walker | return | Hurt |
Closest papers by meaning
- Predictive Coding Enhances Meta-RL To Achieve Interpretable Bayes-Optimal Belief Representation Under Partial Observability (NeurIPS 2025 · similarity 0.80)
- Contrastive Difference Predictive Coding (ICLR 2024 · similarity 0.79)
- How Patterns Dictate Learnability in Sequential Data (NeurIPS 2025 · similarity 0.79)
- Predictive auxiliary objectives in deep RL mimic learning in the brain (ICLR 2024 · similarity 0.79)
- Predictive Attractor Models (NeurIPS 2024 · similarity 0.78)
- CogDPM: Diffusion Probabilistic Models via Cognitive Predictive Coding (ICML 2024 · similarity 0.78)
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.