Research map / Reinforcement learning / Policy gradient and actor-critic
The Nature of Temporal Difference Errors in Multi-step Distributional Reinforcement Learning
Yunhao Tang, Remi Munos, Mark Rowland, Bernardo Avila Pires, Will Dabney, Marc G Bellemare
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 |
|---|---|---|---|
| increase trace cutoff c-bar in distributional Retrace (ablation) | Distributional Retrace (quantile) · Tabular MDPs | fixed point quality L_p | Improved |
| use uncorrected n-step instead of Retrace (ablation) | multi-step distributional RL · Tabular MDPs | fixed point bias | Hurt |
| multi-step Retrace (n=3) instead of one-step | C51 · Atari-57 | median human-normalized score | Improved |
| Retrace instead of uncorrected n-step | C51 · Atari-57 | median human-normalized score | Improved |
| multi-step variants (Retrace, n-step) instead of one-step | QR-DQN · Atari-57 | mean and median human-normalized score | Improved |
| Retrace instead of uncorrected n-step | QR-DQN · Atari-57 | mean human-normalized score | Improved |
| Retrace instead of uncorrected n-step | QR-DQN · Atari-57 | median human-normalized score | No clear effect |
Closest papers by meaning
- Distributional Reinforcement Learning With Quantile Regression (AAAI 2018 · similarity 0.90)
- Action Gaps and Advantages in Continuous-Time Distributional Reinforcement Learning (NeurIPS 2024 · similarity 0.88)
- A Perspective of Q-value Estimation on Offline-to-Online Reinforcement Learning (AAAI 2024 · similarity 0.88)
- ODICE: Revealing the Mystery of Distribution Correction Estimation via Orthogonal-gradient Update (ICLR 2024 · similarity 0.88)
- Categorical Distributional Reinforcement Learning with Kullback-Leibler Divergence: Convergence and Asymptotics (ICML 2025 · similarity 0.87)
- Intrinsic Benefits of Categorical Distributional Loss: Uncertainty-aware Regularized Exploration in Reinforcement Learning (NeurIPS 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.
