Research map / Reinforcement learning / Policy gradient and actor-critic
Off-Policy Deep Reinforcement Learning by Bootstrapping the Covariate Shift
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 discounted COP-TD corrections instead of no corrections | C51 · Pong | performance | Helped |
| Use a discount factor of 0.9999 for COP-TD (ablation) | C51 · Pong | performance | Hurt |
| Add ratio normalization (ablation) | C51 with COP-TD · Pong | performance | Helped |
| Use a high normalization weight (ablation) | C51 with COP-TD normalization · Pong | stability | Hurt |
| Use discounted COP-TD corrections with γ̂ = 0.99 | C51 · Seaquest | performance | Helped |
| Use discounted COP-TD corrections with γ̂ = 0.99 | C51 · Breakout | performance | Helped |
| Use discounted COP-TD corrections with γ̂ = 0.99 | C51 · Pong | performance | Helped |
| Use discounted COP-TD corrections with γ̂ = 0.99 | C51 · Asterix | performance | No clear effect |
| Use discounted COP-TD corrections with γ̂ = 0.99 | C51 · Space Invaders | performance | Hurt |
| Train the ratio model as an auxiliary task without using the covariate-shift prediction for correction (ablation) | C51 · Atari games | performance | No clear effect |
| Use TD-error-based prioritized sampling instead of covariate-shift prioritization (ablation) | C51 · Atari games | performance | No clear effect |
| Use the covariate-shift prediction as an auxiliary task with uniform sampling (ablation) | C51 · Space Invaders | performance | Helped |
Closest papers by meaning
- Chaining Value Functions for Off-Policy Learning (AAAI 2022 · similarity 0.90)
- Interpolated Policy Gradient: Merging On-Policy and Off-Policy Gradient Estimation for Deep Reinforcement Learning (NeurIPS 2017 · similarity 0.89)
- Relaxed Stationary Distribution Correction Estimation for Improved Offline Policy Optimization (AAAI 2024 · similarity 0.87)
- Efficient Off-Policy Learning for High-Dimensional Action Spaces (ICLR 2025 · similarity 0.87)
- A Unifying View of Linear Function Approximation in Off-Policy RL Through Matrix Splitting and Preconditioning (NeurIPS 2025 · similarity 0.87)
- Iteratively Refined Behavior Regularization for Offline Reinforcement Learning (NeurIPS 2024 · 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.