Research map / Reinforcement learning / Policy gradient and actor-critic
Distributional value gradients for stochastic environments
Baptiste Debes, Tinne Tuytelaars
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| model joint return and gradient distribution with MSMMD/MMD Sobolev critic | deterministic and gradient-agnostic critics · 2D point-mass toy task, growing number of bonus locations | success (reward reached) | Improved |
| use deterministic Huber Sobolev (MAGE) critic | gradient-agnostic critics · 2D point-mass toy task | success (reward reached) | No clear effect |
| use provably contractive MSMMD instead of plain MMD (ablation) | DSDPG · 2D point-mass toy task | success (reward reached) | Improved |
| DSDPG in noise-free MuJoCo | baselines (TD3-Huber, MAGE, IQN, MMD) · six MuJoCo tasks, no noise | final return | No clear effect |
| DSDPG under multiplicative observation noise | baselines · six MuJoCo tasks, observation noise | final return | Mixed |
| DSDPG under additive Gaussian dynamics noise | baselines · six MuJoCo tasks, Gaussian dynamics noise | final return | Mixed |
| deterministic Sobolev (MAGE) under observation noise | MAGE · Walker2d-v2 and Humanoid-v2, observation noise | final return | Hurt |
| add overestimation-bias correction (TQC truncation / double estimation) (ablation) | DSDPG and deterministic baselines · MuJoCo | return | Improved |
| replace cVAE world model with normalizing flow (ablation) | DSDPG · MuJoCo tasks | return | No clear effect |
Closest papers by meaning
- Bayesian Distributional Policy Gradients (AAAI 2021 · similarity 0.90)
- Near-Minimax-Optimal Distributional Reinforcement Learning with a Generative Model (NeurIPS 2024 · similarity 0.89)
- Quotient-Categorical Representations for Bellman-Compatible Average-Reward Distributional Reinforcement Learning (NeurIPS 2026 · similarity 0.88)
- Intrinsic Benefits of Categorical Distributional Loss: Uncertainty-aware Regularized Exploration in Reinforcement Learning (NeurIPS 2025 · similarity 0.88)
- Bellman Unbiasedness: Toward Provably Efficient Distributional Reinforcement Learning with General Value Function Approximation (ICML 2025 · similarity 0.88)
- Foundations of Multivariate Distributional 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.
