I am currently a PostDoc at Inria Lille in the Scool team. I am under the supervision Emilie Kaufmann and Remy Degenne. My research interests are around sequential learning, mostly for bandits (Active identification and cumulative regret) and communication / information theory.
Between 2022 and 2025, I was a PhD Student at the L2S lab at CentraleSupélec under the supervision of Richard Combes and Sheng Yang
We show that the lower bound for Linear Bandit for ellipsoidal action set is $\Omega(\min(d \sigma \sqrt{T} + d \|\theta\|_{A}, \|\theta\|_{A} T))$ and that for any algorithm hard …
We propose the first known Thompson Sampling for combinatorial bandits whose finite-time regret does not scale exponentially with the dimension of the problem. Suprisingly, …
Thompson Sampling (TS) for combinatorial semi-bandits is sub-optimal in the sense that its regret scales exponentially in the ambient dimension for some well chosen exemples.