François-Pierre Paty

I am a research scientist at Ida, a French AI startup. I am also the founder of HSK level, a Chinese learning app.
Prior to that, I worked as a postdoctoral researcher at CREST where I developped algorithms for economic models.
I defended my PhD in applied mathematics and machine learning in June 2021, where I worked under the supervision of Prof. Marco Cuturi.

News

Upcoming events:

Recent events:

October 2, 2023: I joined Ida as a research scientist.
June 27, 2022: I gave a talk in the New Monge Problems seminar at the Université Gustave Eiffel near Paris.
May 11, 2022: I gave a talk (slides in .pdf) at the Mokaplan team seminar at INRIA in Paris.
October 21–22, 2021: I participated in the Paris workshop on optimal transport with applications to economics and statistics at Sciences Po Paris.
June 29, 2021: I defended my PhD thesis (slides in pdf and the manuscript) at ENSAE Paris.
March 31, 2021: I gave a talk at the EDMH PhD students seminar (video in French).
March 4, 2021: I gave a talk at the Image, Optimization and Probability seminar at the Institut de Mathématiques de Bordeaux.

Older events (click to reveal):

August 28, 2020: We received a Notable Paper Award at AISTATS 2020 for our paper "Regularity as Regularization: Smooth and Strongly Convex Brenier Potentials in Optimal Transport".
August 26–28, 2020: I presented our paper "Regularity as Regularization: Smooth and Strongly Convex Brenier Potentials in Optimal Transport" at AISTATS 2020.
July 12–18, 2020: I presented our paper "Regularized Optimal Transport is Ground Cost Adversarial" at ICML 2020.
January 17, 2020: I gave a talk (.key, .pdf) at the seminar day Learning meets Astrophysics at CEA in Saclay, France.
December 08–14, 2019: I presented our paper Regularity as Regularization: Smooth and Strongly Convex Brenier Potentials at NeurIPS Optimal Transport and Machine Learning Workshop in Vancouver, Canada.
November 20, 2019: I gave an introductory lecture on optimal transport at the seminar Stat·Eco·ML in Palaiseau, France.
November 05, 2019: I gave a talk (.key, .pdf) at Le Séminaire Palaisien in Palaiseau, France.
August 26–31, 2019: I participated in, and gave a tutorial (notebooks, video) at Machine Learning Summer School in Moscow, Russian Federation.
July 07–19, 2019: I participated in, gave a talk (.pdf) and presented a poster (.pdf), in Saint-Flour Probability Summer School in Saint-Flour, France.
June 24–28, 2019: I participated in the workshop People in Optimal Transportation and Applications in Cortona, Italy.
June 09–15, 2019: I presented (poster and 20-minute oral) our paper "Subspace Robust Wasserstein Distances" at ICML 2019 in Long Beach, USA.
March 25–29, 2019: I presented a poster (.pdf) at the workshop Optimization and Statistical Learning in Les Houches, France.

Research interests

I've been working on algorithmic optimal transport (OT), with applications to machine learning and economics. My work focuses on designing algorithms for OT problems as well as OT tools that are more robust to the curse of dimensionality, to data corruption or noise, so that OT can be efficiently applied to real data problems.

Introduction au transport optimal (en français)

Voici une introduction au transport optimal extraite d'un séminaire de vulgarisation que j'ai donné à destination des doctorants de l'École doctorale Hadamard en mars 2021.

Publications

My PhD manuscript can be found here.

Authors	Paper	Conference	Resources
F-P. Paty P. Choné F. Kramarz	Algorithms for Weak Optimal Transport with an Application to Economics	Preprint 2022	[ArXiv]
F-P. Paty M. Cuturi	Regularized Optimal Transport is Ground Cost Adversarial	ICML 2020	[Proceedings][ArXiv] [Slides] [Video]
F-P. Paty A. d'Aspremont M. Cuturi	Regularity as Regularization: Smooth and Strongly Convex Brenier Potentials in Optimal Transport	AISTATS 2020 (Notable Paper Award)	[Proceedings][ArXiv] [Slides] [Video]
F-P. Paty M. Cuturi	Subspace Robust Wasserstein Distances	ICML 2019 (Top 20%)	[Proceedings] [ArXiv] [Code] [Poster] [Video]

Talks & Tutorials

June 27, 2022: I gave a talk (on the blackboard) in the New Monge Problems seminar at the Université Gustave Eiffel near Paris.
May 11, 2022: I gave a talk (slides in .pdf) at the Mokaplan team seminar at INRIA in Paris.
June 29, 2021: I defended my PhD (slides in pdf and the manuscript) at ENSAE Paris.
March 31, 2021: I gave a talk (video in French) at the EDMH PhD students seminar.
March 4, 2021: I gave a talk at the Image, Optimization and Probability seminar at the Institut de Mathématiques de Bordeaux.
August 2020: I gave a talk (.pdf) at AISTATS 2020.
July 2020: I gave a talk (.pdf) at ICML 2020.
January 2020: I gave a talk (.key, .pdf) at the seminar day Learning meets Astrophysics.
November 2019: I gave a talk (on the blackboard) at the seminar Stat·Eco·ML.
November 2019: I gave a talk (.key, .pdf) at Le Séminaire Palaisien.
August 2019: I gave a tutorial (notebooks, video) at the Machine Learning Summer School 2019 in Moscow.
July 2019: I gave a talk (.pdf) at Saint-Flour Probability Summer School.
June 2019: I gave a 20-minute oral presentation (video) at ICML.

Education

PhD from Institut Polytechnique de Paris
Engineering degree from Ecole Polytechnique
Engineering degree from ENSAE Paris
Masters degree in Statistics and Machine Learning from Université Paris-Sud

Teaching

2020-2021:

Topologie et Analyse Fonctionnelle, ENSAE 1st year students, Fall 2020, Prof.: Laurent Decreusefond
---> Correction des exercices (in French)
Statistique Mathématique, ENSAE 3rd year students, Fall 2020, Prof.: Vincent Cottet
Optimisation différentiable, ENSAE 1st year students, Spring 2021, Prof.: Guillaume Lecué

2019-2020:

Optimal Transport: Theory, Computations, Statistics, and ML Applications, ENSAE 3rd year students, Spring 2020, Prof.: Marco Cuturi
Deep Learning: Models and Optimization, ENSAE 3rd year students, Spring 2020, Prof.: Marco Cuturi
Optimisation différentiable, ENSAE 1st year students, Spring 2020, Prof.: Guillaume Lecué
---> Correction des exercices (in French)
Topologie et Analyse, ENSAE 1st year students, Fall 2019, Prof.: Nicolas Marie

2018-2019:

Geometric Methods in Machine Learning, ENSAE 3rd year students, Spring 2019, Prof.: Marco Cuturi
---> Jupyter Notebooks
Stochastic Optimization and Automatic Differentiation for Machine Learning, ENSAE 3rd year students, Spring 2019, Prof.: Marco Cuturi
---> Jupyter Notebooks
Optimisation différentiable, ENSAE 1st year students, Spring 2019, Prof.: Guillaume Lecué
Topologie et Analyse, ENSAE 1st year students, Fall 2018, Prof.: Nicolas Marie