Research map / Reinforcement learning / Policy gradient and actor-critic
A Temporal-Difference Approach to Policy Gradient Estimation
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 |
|---|---|---|---|
| Lower trace parameter lambda (more gradient critic, less semi-gradient) in LSTDΓ (ablation) | LSTDΓ policy gradient · Imani's MDP | bias and variance of gradient estimate | Helped |
| Lower lambda (weak mixing of gradient critic) in LSTDΓ (ablation) | LSTDΓ policy gradient · Imani's MDP | final performance | Helped |
| Lower lambda in TDRCΓ (gradient critic usage) (ablation) | TDRCΓ · Randomly generated MDPs (30 states, 2 actions) | return | Helped |
| Apply gradient critic only to last actor layer (TDRCΓ-LastLayer) vs all parameters (ablation) | TDRCΓ · Randomly generated MDPs | return | No clear effect |
| Gradient critic (TDRCΓ) vs OffPAC and ACE(1) baselines | TDRCΓ · Randomly generated MDPs | return | Helped |
| Sharing features between value and gradient critic | gradient critic · theory | unbiasedness of gradient critic | Helped |
Closest papers by meaning
- Correcting discount-factor mismatch in on-policy policy gradient methods (ICML 2023 · similarity 0.91)
- Statistically Efficient Off-Policy Policy Gradients (ICML 2020 · similarity 0.90)
- PAGE-PG: A Simple and Loopless Variance-Reduced Policy Gradient Method with Probabilistic Gradient Estimation (ICML 2022 · similarity 0.90)
- Action-Level Behavior Policy Optimization for Variance-Reduced Policy Gradients (NeurIPS 2026 · similarity 0.90)
- A Parametric Class of Approximate Gradient Updates for Policy Optimization (ICML 2022 · similarity 0.89)
- Reusing Trajectories in Policy Gradients Enables Fast Convergence (ICML 2026 · similarity 0.89)
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.