Postdoc at @NYU_Courant, AI for Science PhD from @ENS_ULM and @Inria
Joined November 2018
- Tweets8
- Following313
- Followers105
- Likes82
I'm excited to present CASAL, our training-free constrained sampling algorithm, at #ICLR2026! 🌴
Leveraging Wasserstein optimization, CASAL enforces hard constraints on pre-trained diffusion models with minimal bias.
Saturday 10:30 AM – 1:00 PM, P3-#409
arxiv.org/abs/2505.18017
Work done with @YongquanQu2340, Sara Shamekh and @PierreGentine at @NYU_Courant and at the @LeapStc center, in collaboration with @M2LInES
Tomorrow, I will present our paper Interpretable Meta-Learning of Physical systems at #ICLR2024, poster #29, 4.30 P.M.
We propose CAMEL, a meta-learning algorithm incorporating physical structure. Applications include physical parameter identification and robotics.
@marc_lelarge
Matthieu Blanke retweeted
Hi! I am not in Neurips at NO, but very happy to share our poster with you! @pesme_scott either, but if you are interested please talk to @suriyagnskr or Nicolas (@tml_lab) who are present !
Matthieu Blanke retweeted
Today, Matthieu Blanke @MatB_29 from @inria_paris presents his work "Interpretable Meta-Learning of Physical Systems" during the team meeting!
Although I decided not to fly to Hawaii for #ICML2023, our poster on exploration in reinforcement learning will be displayed on Wednesday: session 4, poster 544. 🌱🚫✈️
@marc_lelarge
mb-29.github.io/projects/exp…
Interested? Let's chat online!
Our work "FLEX: an Adaptive Exploration Algorithm for Nonlinear Systems" has been accepted at #ICML2023 ! @marc_lelarge
FLEX is a lightweight model-based pure exploration policy maximizing the Fisher information.
Paper: arxiv.org/abs/2304.13426
Full demo: youtube.com/watch?v=hGpkdz8-…