Clément Bonet
@Clement_Bonet_
Followers
232
Following
470
Media
10
Statuses
42
Assistant Professor at École Polytechnique interested in Optimal Transport.
Paris, France
Joined January 2021
#Distinction 🏆| Charlotte Pelletier, lauréate d'une chaire @InstUnivFr, développe des méthodes d’intelligence artificielle appliquées aux séries temporelles d’images satellitaires. ➡️ https://t.co/yTiYJy2LIJ 🤝 @irisa_lab @CNRS_dr17
0
1
3
I'm thrilled to announce that my #ERCStG project **Optinfinite : Efficient infinite-dimensional optimization over measures** has been accepted. Thank you @ERC_Research ! Many thanks also to @CrestUmr @IP_Paris_ for their support, as well as to my collaborators and friends.
11
7
75
In Vancouver for #ICML2025! I'll present our work during oral session 1D at 10:45 tomorrow (Ballroom C), followed by poster session E2111. We extend conservation laws from simplified settings to real-world architectures like Transformers and ResNets, then study them under SGD.
1
3
13
With Christophe, we will present our work tuesday. 📍Oral: West Ballroom D, Poster: East Exhibition Hall A-B #E-1300 📅 Tuesday, July 15th, 4 p.m. for the Oral, and between 4:30 p.m. and 7 p.m for the Poster. See you there!
0
1
4
🎉 Happy to share that our work "Flowing Datasets with Wasserstein over Wasserstein Gradient Flows" was accepted at #ICML2025 as an oral! This is a joint work with the amazing Christophe Vauthier and @Korba_Anna! Link: https://t.co/4aJ0DkKdCC
1
24
101
Clément Bonet (ENSAE/CREST), Flowing Datasets with Wasserstein over Wasserstein Gradient Flows
0
1
3
With Christophe, we will present our work tuesday. 📍Oral: West Ballroom D, Poster: East Exhibition Hall A-B #E-1300 📅 Tuesday, July 15th, 4 p.m. for the Oral, and between 4:30 p.m. and 7 p.m for the Poster. See you there!
0
1
4
We apply this scheme to minimize the MMD with kernels based on the Sliced-Wasserstein distance. And as applications, we flow dataset of images to solve tasks such as transfer learning and dataset distillation.
1
0
2
We leverage this gradient to do optimization over this space. We update each particle using this gradient, and observe several layers of interactions, between the particles and between the classes.
1
0
0
To solve this task, we endow the space with the Wasserstein over Wasserstein (WoW) distance, and exploit its Riemannian structure. It gives us a way to define a notion of gradient.
1
0
0
In our work, we propose to model labeled datasets as probability over probability distributions, and to frame the task of flowing datasets as a minimization of a discrepancy over this space.
2
0
1
🎉 Happy to share that our work "Flowing Datasets with Wasserstein over Wasserstein Gradient Flows" was accepted at #ICML2025 as an oral! This is a joint work with the amazing Christophe Vauthier and @Korba_Anna! Link: https://t.co/4aJ0DkKdCC
1
24
101
Check out this great work led by @HanlinYu2025 and follow him! We learn the time score faster and more accurately (in pixel space too). And use it to estimate density ratios, energy-based models, and generate samples.
Happy to announce our ICML 2025 paper with Arto Klami, Aapo Hyvärinen, @Korba_Anna and @lomarchehab! Density Ratio Estimation is hard, especially in high dimensions! This was reformulated as learning and integrating a time score. We improve the learning! 🧵 Thread below
0
2
14
If you're at #AISTATS2025, check out the presentation by Jonathan Geuter, in collaboration with @Clement_Bonet_ , @Korba_Anna and @elmelis. 'DDEQs: Distributional Deep Equilibrium Models through Wasserstein Gradient Flows' https://t.co/Aac1kwe2lC
#AI #statistics #ML
0
7
18
'Sliced-Wasserstein Distances and Flows on Cartan-Hadamard Manifolds', by Clément Bonet, Lucas Drumetz, Nicolas Courty. https://t.co/zcMboLYgJc
#wasserstein #manifolds #manifold
0
3
16
Slicing Unbalanced Optimal Transport Clément Bonet, Kimia Nadjahi, Thibault Sejourne, Kilian FATRAS, Nicolas Courty. Action editor: Benjamin Guedj. https://t.co/NgWrCUPg38
#outliers #transport #optimal
openreview.net
Optimal transport (OT) is a powerful framework to compare probability measures, a fundamental task in many statistical and machine learning problems. Substantial advances have been made in...
0
2
14
Today something crazy happened. POT has reached 1000 citations (total) 🤩🚀. Very proud to be part of a scientific community that acknowledges open source research software. Please continue to use, cite and contribute to POT ! Small🧵below for those interested
2
20
139
From Diffusion Models to Schrödinger Bridges - A shame not to see this NeurIPS keynote live, by the incredible @ArnaudDoucet1 - SB naturally extends flow/ bridge matching and diffusion models - Particularly useful for data to data - Links to OT https://t.co/IIpuU0Nw8o
0
23
147
Excited about Wasserstein gradient flows? We extend mirror and preconditioned gradient descent on the space of probability distributions. Join Clément and me tomorrow (Thu 12th) at 11:00 AM, poster #5911. Looking forward to great discussions! 🌊 @Clement_Bonet_ @Korba_Anna
1
8
69