Research map / Reinforcement learning / Policy gradient and actor-critic
A Parametric Class of Approximate Gradient Updates for Policy Optimization
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 MLA scaling instead of the other evaluated scaling functions | MSE, MVE, and PGPB updates · Synthetic 2D contextual bandit | convergence speed and final objective value | Helped |
| Use MLA scaling with PGPB form rather than the other evaluated update combinations | Policy optimization updates · Synthetic 2D contextual bandit | regret | Helped |
| Use nonzero α_r MLA scaling instead of identity scaling (ablation) | Policy gradient · FourRoom | return | Helped |
| Use nonzero α_r MLA scaling instead of identity scaling (ablation) | Q-learning · FourRoom | return | Helped |
| Vary α_r away from zero (ablation) | MLA-scaled policy gradient and Q-learning · FourRoom | learning speed | Helped |
| Replace PPO's policy-gradient scale with parameterized MLA scaling (ablation) | PPO · MuJoCo | performance | Helped |
Closest papers by meaning
- $\phi$-Update: A Class of Policy Update Methods with Policy Convergence Guarantee (ICLR 2025 · similarity 0.91)
- Improving Value Estimation Critically Enhances Vanilla Policy Gradient (ICML 2025 · similarity 0.90)
- A Closer Look at Deep Policy Gradients (ICLR 2020 · similarity 0.90)
- Convergence Analysis of Policy Gradient Methods with Dynamic Stochasticity (ICML 2025 · similarity 0.88)
- Natural Policy Gradient as Doubly Smoothed Policy Iteration: A Bellman-Operator Framework (NeurIPS 2026 · similarity 0.88)
- Adaptive-Gradient Policy Optimization: Enhancing Policy Learning in Non-Smooth Differentiable Simulations (ICML 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.