Research map / Reinforcement learning / Policy gradient and actor-critic
Deterministic Policy Optimization by Combining Pathwise and Score Function Estimators for Discrete Action Spaces
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 |
|---|---|---|---|
| Use RPG instead of A3C | policy optimization · Cart Pole | samples until solve: $279$ -> $91$ | Helped |
| Use RPG instead of CEM | policy optimization · Cart Pole | samples until solve: $470$ -> $91$ | Helped |
| Use RPG instead of A3C | policy optimization · Acrobot | samples until solve: $541$ -> $311$ | Helped |
| Use RPG instead of CEM | policy optimization · Acrobot | samples until solve: $180$ -> $311$ | Hurt |
| Use RPG instead of A3C | policy optimization · Hand Mass | mean reward: $-2.42 \pm 2.19$ -> $-0.026 \pm 0.01$ | Helped |
| Use RPG instead of CEM | policy optimization · Hand Mass | mean reward: $-0.086 \pm 0.01$ -> $-0.026 \pm 0.01$ | Helped |
| Use RPG instead of A3C | policy optimization · Mountain Car | mean reward: $-176 \pm 38.6$ -> $-131.3 \pm 5.3$ | Helped |
| Use RPG instead of CEM | policy optimization · Mountain Car | mean reward: $ -110.3 \pm 1.2$ -> $-131.3 \pm 5.3$ | Hurt |
| Train with a smooth surrogate reward rather than the true reward | RPG · classical control tasks except Hand Mass | actual reward | No clear effect |
| Train A3C with smoothed rewards rather than real rewards | A3C · classical control tasks | performance | No clear effect |
Closest papers by meaning
- Truly Deterministic Policy Optimization (NeurIPS 2022 · similarity 0.91)
- Ranking Policy Gradient (ICLR 2020 · similarity 0.89)
- Score Regularized Policy Optimization through Diffusion Behavior (ICLR 2024 · similarity 0.88)
- Learning Optimal Deterministic Policies with Stochastic Policy Gradients (ICML 2024 · similarity 0.88)
- Actor-Critics Can Achieve Optimal Sample Efficiency (ICML 2025 · similarity 0.88)
- Diffusion Policy Policy Optimization (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.