Research map / Reinforcement learning / Policy gradient and actor-critic
Distributional Reinforcement Learning With Quantile Regression
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 |
|---|---|---|---|
| Replace C51's fixed support and projection with quantile locations trained using Huber quantile loss | DQN · Atari 2600, 57 games (best agent) | Mean human-normalized score: 701% -> 915% | Helped |
| Replace C51's fixed support and projection with quantile locations trained using Huber quantile loss | DQN · Atari 2600, 57 games (best agent) | Median human-normalized score: 178% -> 211% | Helped |
| Use Huber quantile loss (κ = 1) instead of strict quantile loss (κ = 0) (ablation) | QR-DQN · Atari 2600, 57 games (best agent) | Mean human-normalized score: 881% -> 915% | Helped |
| Use Huber quantile loss (κ = 1) instead of strict quantile loss (κ = 0) (ablation) | QR-DQN · Atari 2600, 57 games (best agent) | Median human-normalized score: 199% -> 211% | Helped |
| Use distributional quantile regression instead of value-only TD learning | QR-TD · Two-room windy gridworld | 1-Wasserstein value-distribution error | Helped |
| Use distributional quantile regression instead of value-only TD learning | QR-TD · Two-room windy gridworld | Squared value-function error | No clear effect |
Closest papers by meaning
- Fully Parameterized Quantile Function for Distributional Reinforcement Learning (NeurIPS 2019 · similarity 0.94)
- Implicit Quantile Networks for Distributional Reinforcement Learning (ICML 2018 · similarity 0.93)
- A Distributional Perspective on Reinforcement Learning (ICML 2017 · similarity 0.93)
- Quantile Geometry Regularization for Distributional Reinforcement Learning (NeurIPS 2026 · similarity 0.89)
- Near-Minimax-Optimal Distributional Reinforcement Learning with a Generative Model (NeurIPS 2024 · similarity 0.89)
- A Distributional Analogue to the Successor Representation (ICML 2024 · 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.