Research map / Reinforcement learning / Visual and representation RL
On the Importance of Exploration for Generalization in Reinforcement Learning
Yiding Jiang, J Zico Kolter, Roberta Raileanu
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 TEE (training on diverse data) to QR-DQN without UCB (ablation) | QR-DQN · Procgen | normalized score | Improved |
| add epistemic-uncertainty UCB exploration to QR-DQN (ablation) | QR-DQN · Procgen | normalized score | Improved |
| add TEE on top of QR-DQN+UCB (full method) (ablation) | QR-DQN+UCB · Procgen | normalized score | Improved |
| explore with total uncertainty instead of epistemic uncertainty (ablation) | QR-DQN+UCB · Procgen | test score | Hurt |
| use Bootstrapped DQN or DQN+UCB (total uncertainty) instead of plain DQN | DQN · Procgen | normalized score | Hurt |
| use epsilon-z-greedy exploration on QR-DQN | QR-DQN · Procgen | normalized score | No clear effect |
| use UA-DQN instead of deep ensemble UCB | QR-DQN+UCB · Procgen | normalized score | Hurt |
| use NoisyNet exploration instead of EDE | QR-DQN · Procgen | normalized score | Hurt |
| use EDE instead of QR-DQN | Procgen agent · Procgen | median score: almost 3 times the median score | Improved |
| use EDE instead of QR-DQN | Procgen agent · Procgen | IQM: more than 2 times the IQM | Improved |
| use EDE instead of Rainbow exploration | Rainbow · Crafter | score (%): 4.3 -> 11.7 | Improved |
| use EDE instead of DreamerV2 | Crafter agent · Crafter | score (%): 10.0 -> 11.7 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Rainbow — Rainbow: Combining Improvements in Deep Reinforcement Learning (AAAI 2018) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Zero-Shot Context Generalization in Reinforcement Learning from Few Training Contexts (NeurIPS 2025 · similarity 0.88)
- Diverse Projection Ensembles for Distributional Reinforcement Learning (ICLR 2024 · similarity 0.88)
- Random Latent Exploration for Deep Reinforcement Learning (ICML 2024 · similarity 0.87)
- Intrinsic Benefits of Categorical Distributional Loss: Uncertainty-aware Regularized Exploration in Reinforcement Learning (NeurIPS 2025 · similarity 0.87)
- Off-policy Reinforcement Learning with Model-based Exploration Augmentation (NeurIPS 2025 · similarity 0.87)
- On Generalization Across Environments In Multi-Objective Reinforcement Learning (ICLR 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.
