Distributional Value Gradients for Stochastic Environments

Plain-English Summary

A distributional value-gradient approach for improving value learning in stochastic or noisy environments.

Abstract

Extends distributional reinforcement learning on continuous state-action spaces to model distributions over scalar value functions and their gradients for stochastic environments.

@inproceedings{debes2026distributionalvaluegradients,
  title = {Distributional Value Gradients for Stochastic Environments},
  author = {Debes, Baptiste and Tuytelaars, Tinne},
  booktitle = {International Conference on Learning Representations},
  year = {2026},
  url = {https://arxiv.org/abs/2601.20071},
  doi = {10.48550/arXiv.2601.20071}
}

Related Posts