Research map / Reinforcement learning / Policy gradient and actor-critic
Statistically Efficient Off-Policy Policy Gradients
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 EOPPG instead of existing off-policy policy-gradient estimators | Off-policy policy gradient in MDP · MDP | MSE: \Theta(H^{4}/n) versus 2^{\Theta(H)}\Theta(1/n) | Helped |
| Use EOPPG with MDP structure rather than NMDP structure | EOPPG · Off-policy policy gradient | MSE horizon dependence: \Theta(H^{4}/n) versus 2^{\Theta(H)}\Theta(1/n) | Mixed |
| Corrupt the (q, d^q) nuisance models with noise (ablation) | EOPPG · Simple environment, gradient estimation | MSE | No clear effect |
| Corrupt the (mu, d^mu) nuisance models with noise (ablation) | EOPPG · Simple environment, gradient estimation | MSE | No clear effect |
| Corrupt the (d^mu, d^q) nuisance models with noise (ablation) | EOPPG · Simple environment, gradient estimation | MSE | Mixed |
| Use EOPPG for gradient ascent rather than REINFORCE or PG | Off-policy policy learning · Simple environment, final-policy regret | regret | Helped |
Closest papers by meaning
- Doubly Robust Off-Policy Value and Gradient Estimation for Deterministic Policies (NeurIPS 2020 · similarity 0.92)
- Interpolated Policy Gradient: Merging On-Policy and Off-Policy Gradient Estimation for Deep Reinforcement Learning (NeurIPS 2017 · similarity 0.92)
- Robust On-Policy Sampling for Data-Efficient Policy Evaluation in Reinforcement Learning (NeurIPS 2022 · similarity 0.92)
- Unifying Gradient Estimators for Meta-Reinforcement Learning via Off-Policy Evaluation (NeurIPS 2021 · similarity 0.92)
- More Efficient Off-Policy Evaluation through Regularized Targeted Learning (ICML 2019 · similarity 0.91)
- Policy Gradient for Rectangular Robust Markov Decision Processes (NeurIPS 2023 · similarity 0.91)
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.