Research map / Reinforcement learning / Policy gradient and actor-critic
On the Hidden Biases of Policy Mirror Ascent in Continuous 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 SRMA with a Cauchy policy instead of Gaussian-policy comparators | Policy search on Pathological Mountain Car · Pathological Mountain Car | cumulative return | Helped |
| Replace the Gaussian policy with a Cauchy policy while retaining RPG updates (ablation) | RPG · Pathological Mountain Car | reward stability | Hurt |
| Use mirror-ascent updates to stabilize Cauchy-policy optimization (ablation) | Cauchy-policy policy search · Pathological Mountain Car | reward stability | Helped |
| Use Cauchy rather than Gaussian policy parameterization | Policy gradients in continuous action spaces · Continuous action spaces | hidden bias | Helped |
| Use recursive gradient tracking with mirror ascent and constant-size batches | SRMA · General non-convex objectives | sample complexity: \mathcal{O}(1) samples per iteration; \mathcal{O}(1/\epsilon^2) sample complexity | Helped |
Closest papers by meaning
- Global Optimality for Constrained Exploration via Penalty Regularization (NeurIPS 2026 · similarity 0.87)
- Policy Mirror Descent with Lookahead (NeurIPS 2024 · similarity 0.87)
- Excluding the Irrelevant: Focusing Reinforcement Learning through Continuous Action Masking (NeurIPS 2024 · similarity 0.87)
- Learning mirror maps in policy mirror descent (ICLR 2025 · similarity 0.87)
- Natural Policy Gradient as Doubly Smoothed Policy Iteration: A Bellman-Operator Framework (NeurIPS 2026 · similarity 0.87)
- Parameter Exploration for RLVR via Variational Learning (NeurIPS 2026 · 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.