Research map / Optimization / Nonconvex and smooth optimization
Convergence and Optimality of Policy Gradient Methods in Weakly Smooth Settings
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 |
|---|---|---|---|
| Allow weak Hölder smoothness and an L₂-integrable score instead of requiring strong smoothness and an absolutely bounded gradient | Policy Gradient and Natural Policy Gradient · Continuous-action MDPs satisfying the paper's assumptions | convergence to a stationary point | Helped |
| Use Natural Policy Gradient rather than standard Policy Gradient for the optimality guarantee | Policy optimization under weak smoothness · Weakly smooth policy classes | optimality gap | Helped |
| Use generalized Gaussian policies with heavier tails rather than Gaussian policies | Single-state exploration problem · Exploration bandit | samples needed to discover the correct action | Helped |
Closest papers by meaning
- On the Convergence Rates of Policy Gradient Methods (ICML 2023 · similarity 0.90)
- Convergence Analysis of Policy Gradient Methods with Dynamic Stochasticity (ICML 2025 · similarity 0.90)
- Convergence of Policy Mirror Descent Beyond Compatible Function Approximation (ICML 2025 · similarity 0.90)
- Natural Policy Gradient as Doubly Smoothed Policy Iteration: A Bellman-Operator Framework (NeurIPS 2026 · similarity 0.90)
- On the Second-Order Convergence of Biased Policy Gradient Algorithms (ICML 2024 · similarity 0.89)
- Identifying Policy Gradient Subspaces (ICLR 2024 · 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.