Victor Boone

In Toulouse, ENSEEIHT.

Sometimes I also go to LAAS!

I am Victor Boone, currently Post-Doc at the Institut de Recherche en Informatique de Toulouse, IRIT. My work focuses on Reinforcement Learning, Multi-Armed Bandits, Game Theory and Queuing Theory (more recently).

You can join me via email victor.boone@irit.fr or boonevv@gmail.com.

I am just a curious fellow that tries to explain what I actually see on my computer when I run learning algorithms, by pushing theory to its limits. I am concerned with finite Markov decision processes, stochastic bandits and finite games in normal form. During my PhD, I have been mostly focusing on regret minimization in Markov decision processes. More recently, my focus is shifting towards infinitely large systems topped with learning considerations, such as queuing systems. Stay tuned!

I did my PhD thesis under the supervision of Bruno Gaujal and Panayotis Mertikopoulos and have defended on November 2024. My manuscript is entirely dedicated to regret minimization in Markov decision processes in the undiscounted infinite horizon setting. Check it out here! I’ve even received a prize for it, the prix de thèse UGA 2025! Hurray!

news

Oct 16, 2025	I gave a talk at SOLACE seminars. Big thanks to Thomas Hira and Céline Comte for organizing it!
Jul 5, 2025	I am visiting Urtzi Ayesta at Toulouse and start a Post-Doc there by the end of summer.
Jun 30, 2025	I was at COLT 2025, and met a lot of cool people there!
Jan 25, 2025	I gave a talk to RL Theory Seminars on January 25. I’ve talked in front of a lot of people that I wanted to meet for quite some time!
Jan 5, 2025	Today, I am starting a Post-Doc at Grenoble! In the same team as before. In the same office. So yeah, I know. I am basically finishing my PhD. There are so much things left to do! I’ve had a burst of ideas during the winter break, and I’ve already answered a few open questions at the end of my PhD manuscript. I can’t wait to write them down :-)

latest posts

selected publications

2025

arxiv

Asymptotically optimal regret in communicating Markov decision processes

Victor Boone

May 2025

Bib HTML PDF

@misc{boone_achieving_2025,
  title = {Asymptotically optimal regret in communicating Markov decision processes},
  author = {Boone, Victor},
  year = {2025},
  month = may,
  eprint = {2505.18064},
  archiveprefix = {arXiv},
  primaryclass = {cs.LG},
  url = {https://arxiv.org/abs/2505.18064}
}

2024

PhD thesis

Optimal regrets in Markov decision processes

Victor Boone

Nov 2024

Bib HTML PDF

@thesis{boone_thesis_2024,
  title = {Optimal regrets in Markov decision processes},
  author = {Boone, Victor},
  url = {https://victor-boone.github.io/assets/pdf/VBOONE-manuscript-v1.1.pdf},
  year = {2024},
  month = nov
}

NeurIPS

Achieving Tractable Minimax Optimal Regret in Average Reward MDPs

Victor Boone, and Zihan Zhang

In Proceedings of the 38th International Conference on Neural Information Processing Systems, Nov 2024

_eprint: 2406.01234, publication is WIP

Bib HTML PDF

@inproceedings{boone_achieving_2024,
  title = {Achieving {Tractable} {Minimax} {Optimal} {Regret} in {Average} {Reward} {MDPs}},
  url = {https://arxiv.org/abs/2406.01234},
  author = {Boone, Victor and Zhang, Zihan},
  booktitle = {Proceedings of the 38th International Conference on Neural Information Processing Systems},
  year = {2024},
  note = {\_eprint: 2406.01234, publication is WIP}
}

2023

arxiv

The Sliding Regret in Stochastic Bandits: Discriminating Index and Randomized Policies

Victor Boone

Nov 2023

arXiv:2311.18437 [cs, eess, math, stat]

Bib HTML PDF

@misc{boone_sliding_2023,
  title = {The {Sliding} {Regret} in {Stochastic} {Bandits}: {Discriminating} {Index} and {Randomized} {Policies}},
  publisher = {arXiv},
  author = {Boone, Victor},
  month = nov,
  year = {2023},
  note = {arXiv:2311.18437 [cs, eess, math, stat]},
  keywords = {Computer Science - Machine Learning, Electrical Engineering and Systems Science - Systems and Control, Mathematics - Optimization and Control, Statistics - Machine Learning}
}