Research map / Learning theory and algorithms / Submodular and game algorithms
Self-correcting Q-learning
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 sequential Q-value estimates to correct the maximization bias | Q-learning · Maximization bias example | maximization bias | Helped |
| Use sequential Q-value estimates to correct the maximization bias | Q-learning · 3×3 grid-world, high reward variability | average reward per time step | Helped |
| Use sequential Q-value estimates to correct the maximization bias | Q-learning · 3×3 grid-world, medium reward variability | average reward per time step | Helped |
| Use sequential Q-value estimates to correct the maximization bias | Q-learning · 3×3 grid-world, low reward variability | average reward per time step | No clear effect |
| Use sequential Q-value estimates to correct the maximization bias | Q-learning · Cliff-walking, zero-variance rewards | cumulative reward | Helped |
| Use the self-correcting action-selection rule in a DQN | DQN · Atari 2600 games | score | Helped |
| Set β=1 (ablation) | Self-correcting Q-learning · Maximization bias example | maximization bias | Mixed |
| Set β=2 (ablation) | Self-correcting Q-learning · Maximization bias example | maximization bias | Helped |
| Set β=4 (ablation) | Self-correcting Q-learning · Maximization bias example | maximization bias | Helped |
| Set β=1 (ablation) | Self-correcting Q-learning · 3×3 grid-world | maximal action-value bias | Mixed |
| Set β=2 (ablation) | Self-correcting Q-learning · 3×3 grid-world | maximal action-value bias | Helped |
| Set β=4 (ablation) | Self-correcting Q-learning · 3×3 grid-world | maximal action-value bias | Helped |
Closest papers by meaning
- Maxmin Q-learning: Controlling the Estimation Bias of Q-learning (ICLR 2020 · similarity 0.91)
- Deep Double Q-learning (NeurIPS 2026 · similarity 0.90)
- Regularized Q-learning through Robust Averaging (ICML 2024 · similarity 0.85)
- Long-Horizon Q-Learning: Accurate Value Learning via n-Step Inequalities (NeurIPS 2026 · similarity 0.85)
- ADDQ: Adaptive distributional double Q-learning (ICML 2025 · similarity 0.85)
- Adaptive $Q$-Network: On-the-fly Target Selection for Deep Reinforcement Learning (ICLR 2025 · similarity 0.84)
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.