Raphaël Boige Phd student in Machine Learning RL and Game Theory @ Inria email: raphael.boige [at] inria.fr [Scholar] [Github] [Twitter]

I am a Phd student (expected end in 2027) at Inria. My Phd focuses on Markov-games theory and self-play algorithms. I'm also quite interested in imperfect information games such as Poker. I am currently advised by Bruno Scherrer and Amine Boumaza.

Prior to Inria, I was a researcher for 3 years at InstaDeep. At that time, I participated in multiple deep learning works (1) diversity and robustness of deep RL agents for continuous control, (2) pre-training general-purpose representation networks for RL agents and (3) large language models adapted to genomics modalities (single-cell RNA-seq).

I love Jax and participated in a few jax open-source projects: Jumanji, QDax,pgx and cfrx. I also contributed to PufferLib.

I graduated from Télécom Paris and IP Paris in 2021 with a B.S. and a M. Eng in Computer Science and Applied Mathematics.

I serve regularly as a reviewer for NeurIps, ICML, ICLR, EWRL, RLC.

Publications

• *NEW* The Surprising Effectiveness of Approximate Value Iteration in Self-Play
  R. Boige, A. Boumaza, B. Scherrer
  Preprint (arxiv.org/abs/2609.09094), EWRL oral presentation, 2026. Site.
• AlphaBeta is not as good as you think: a simple random game model for a better analysis of deterministic game-solving algorithms
  R. Boige, A. Boumaza, B. Scherrer
  Neural Information Processing Systems (NeurIPS), 2025
• Pasta: Pretrained action-state transformer agents
  R. Boige*, Y. Flet-Berliac* A. Flajolet, L. Quaedvlieg, G. Richard, T. Pierrot (*equal contribution)
  Reinforcement Learning Conference (RLC), 2024
• Jumanji: a diverse suite of scalable reinforcement learning environments in jax
  C. Bonnet, D. Luo, D. Byrne, S. Surana, S Abramowitz, P. Duckworth, V. Coyette, L. Midgley, E. Tegegn, T. Kalloniatis, O. Mahjoub, M. Macfarlane, A. Smit, N. Grinsztajn, R. Boige, C. Waters, M. Mimouni, U. Mbou Sob, R. de Kock, S. Singh, D. Furelos-Blanco, V. Le, A. Pretorius, A.Laterre
  International Conference on Learning Representations (ICLR), 2024
• Qdax: A library for quality-diversity and population-based algorithms with hardware acceleration
  F. Chalumeau, B. Lim, R. Boige, M. Allard, L. Grillotti, M. Flageat, V. Macé, G. Richard, A. Flajolet, T. Pierrot, A. Cully
  Journal of Machine Learning Research (JMLR), 2024
• Neuroevolution is a competitive alternative to reinforcement learning for skill discovery
  F. Chalumeau*, R. Boige*, B. Lim, V. Macé, M. Allard, A. Flajolet, A. Cully, T. Pierrot (*equal contribution)
  (Spotlight - top 15%) International Conference on Learning Representations (ICLR), 2023.
• The quality-diversity transformer: Generating behavior-conditioned trajectories with decision transformers
  V. Macé, R. Boige, F. Chalumeau, T. Pierrot, G. Richard, N. Perrin-Gilbert
  Proceedings of the Genetic and Evolutionary Computation Conference (GECCO), 2023
• Gradient-informed quality diversity for the illumination of discrete spaces
  R. Boige*, G.Richard*, J. Dona, T. Pierrot, A. Cully (*equal contribution)
  Proceedings of the Genetic and Evolutionary Computation Conference (GECCO), 2023