Baptiste Debes

Speech and Image Processing Department (PSI - ESAT), KU Leuven

I'm a final-year Ph.D. student supervised by Prof. Tinne Tuytelaars. My research focuses on value-gradient methods in reinforcement learning. I use learned environment models to provide richer training signals than scalar rewards or value targets alone. More broadly, I am interested in decoupled training of neural networks, with the goal of reducing the memory footprint of learning and making training less dependent on end-to-end backpropagation.

Baptiste Debes

News

  1. Recognized as a Top Reviewer at NeurIPS 2026.

  2. Attended EWRL 2026 in Lille and presented two papers (ICLR 2026 and ICML 2026). Value gradients Sliced divergences

  3. Accepted at NeurIPS 2026: Bellman Contraction under MMD: A Unified Framework. arXiv Code

  4. Presented the ICML 2026 poster in Seoul. Paper Poster

  5. New blog post on multivariate distributional RL using sliced divergences. Read the post

Show older updates Show fewer updates
  1. Accepted at ICML 2026: Multivariate Distributional Reinforcement Learning Using Sliced Divergences. arXiv Poster

  2. Presented the distributional value-gradients paper at ICLR in Rio de Janeiro. Photos and post

  3. Accepted at ICLR 2026: Distributional Value Gradients for Stochastic Environments. arXiv

Research Interests

Distributional reinforcement learning

Learning distributions over returns, rather than only expected values, to better represent uncertainty in reinforcement learning problems.

Pointers

  • A Distributional Perspective on Reinforcement Learning, Bellemare, Dabney & Munos arXiv
  • Distributional Value Gradients for Stochastic Environments, Debes & Tuytelaars arXiv
Four histograms showing the next-state return distribution, discounting, reward addition, and categorical projection.
Bellemare, Dabney & Munos (2017)

Value-gradient methods

Using learned environment models to estimate gradients of return targets with respect to actions, giving policy learning access to richer local information.

Pointers

  • Learning Continuous Control Policies by Stochastic Value Gradients, Heess et al. arXiv
  • How to Learn a Useful Critic? Model-Based Action-Gradient-Estimator Policy Optimization, D'Oro & Jaśkowski arXiv
  • Distributional Value Gradients for Stochastic Environments, Debes & Tuytelaars arXiv

Decoupled training of neural networks

Training neural-network modules with local or predicted learning signals, aiming to reduce update locking and memory requirements during optimization.

Pointers

  • Decoupled Neural Interfaces using Synthetic Gradients, Jaderberg et al. arXiv
  • Sobolev Training for Neural Networks, Czarnecki et al. arXiv

Selected Publications

Bellman Contraction under MMD: A Unified Framework

Baptiste Debes and Tinne Tuytelaars

Advances in Neural Information Processing Systems (NeurIPS), Sydney, Australia, 2026

First page of Bellman Contraction under MMD: A Unified Framework.

Distributional Value Gradients for Stochastic Environments

Baptiste Debes and Tinne Tuytelaars

The Fourteenth International Conference on Learning Representations (ICLR), Rio de Janeiro, Brazil, 2026

First page of Distributional Value Gradients for Stochastic Environments.

Multivariate Distributional Reinforcement Learning Using Sliced Divergences

Baptiste Debes and Tinne Tuytelaars

Proceedings of the 43rd International Conference on Machine Learning (ICML), Seoul, South Korea, 2026

First page of Multivariate Distributional Reinforcement Learning Using Sliced Divergences.

Recent Posts