MULTIYEAR PROGRAM
CONFERENCE

Meeting in Mathematical Statistics
Rencontres de Statistique Mathématique
Statistical thinking in the age of AI : robustness, fairness and privacy

18 – 22 December 2023

Scientific Committee & Organizing Committee
Comité scientifique & Comité d’organisation

 

Olga Klopp (ESSEC Business School and CREST )
Mohamed Ndaoud (ESSEC Business School and CREST )
Christophe Pouet  (École Centrale de Marseille)
Alexander Rakhlin (MIT)

We plan to dedicate the 2023 – 2025 series of conferences to challenges and emerging topics in the area of mathematical statistics driven by the adventure of artificial intelligence. Tremendous progress has been made in building up powerful machine learning algorithms such as random forests, gradient boosting or neural networks. These models are exceptionally complex and difficult to interpret but offer enormous opportunities in many areas of application going from science, public policies to business. These sophisticated algorithms are often called “black boxes” as they are very hard to analyze. The widespread use of such predictive algorithms raises extremely important questions of replicability, reliability, robustness or privacy protection. The proposed series of conferences is dedicated to new statistical methods built around these black-box algorithms that leverage their power but at the same time guarantee their replicability and reliability.

The first conference of the cycle is dedicated to the potential of the statistical approach in accompanying the development of our data-driven society. Our goal is to encourage collaboration and knowledge sharing between theoretical computer scientists and mathematical statisticians by bringing them together in Luminy. Both communities possess unique visions, skills and expertise. We hope that merging these strengths will advance the field and bring us closer to solving some of the key challenges such as robustness, fairness and privacy of decision-making algorithms.

Nous prévoyons de consacrer la série de conférences 2023 – 2025 aux défis et sujets émergents relatifs au domaine des statistiques mathématiques motivés par l’intelligence artificielle. Des progrès remarquables ont été fait depuis la construction de puissants algorithmes d’apprentissage automatique tels que les forêts aléatoires, le gradient boosting ou les réseaux de neurones. Ces modèles, exceptionnellement complexes et difficiles à interpréter, offrent d’énormes opportunités dans de nombreux domaines d’application allant de la science au business en passant par les politiques publiques. Ces algorithmes sophistiqués sont souvent appelés « boîtes noires » car elles sont très difficiles à analyser. L’utilisation fortement répandue de tels algorithmes prédictifs soulève des questions extrêmement importantes de réplicabilité, de fiabilité, de robustesse ainsi que de protection de la vie privée. Cette série de conférences, que nous proposons, est dédiée aux nouvelles méthodes statistiques construites autour de ces algorithmes « boîtes noires » qui tirent parti de leur puissance tout en garantissant leur réplicabilité et leur fiabilité.

La première conférence de ce cycle est consacrée au potentiel des approches statistiques dans l’accompagnement du développement d’une société axée sur la donnée qu’est la nôtre. Notre objectif est d’encourager la collaboration et le partage des connaissances entre informaticiens théoriciens et statisticiens mathématiciens en les réunissant à Luminy. Les deux communautés possèdent visions, compétences et savoir-faire. Nous espérons que la fusion de ces forces fera progresser le domaine et nous rapprochera de la résolution de certains des problèmes majeurs tels que la robustesse, l’équité et la confidentialité dans la prise de décision des algorithmes.

LECTURES

Abstract: The field of Robust Statistics studies the problem of designing estimators that perform well even when the data significantly deviates from the idealized modeling assumptions. The classical statistical theory, going back to the pioneering works by Tukey and Huber in the 1960s, characterizes the information-theoretic limits of robust estimation for a number of statistical tasks. On the other hand, until fairly recently, the computational aspects of this field were poorly understood. Specifically, no scalable robust estimation methods were known in high dimensions, even for the most basic task of mean estimation.

A recent line of work in computer science developed the first computationally efficient robust estimators in high dimensions for a range of learning tasks. This tutorial will provide an overview of these algorithmic developments and discuss some open problems in the area.

Abstract: I will provide a broad overview of differential privacy, which provides guarantees that a data analysis protects the privacy of data contributors. The main focus will be on the private computation and release of different statistics, both classical (low-dimensional) and high-dimensional statistics. In addition to giving a high-level program for the development of optimal private estimators, I will likely discuss a few open questions as well.

SPEAKERS

Fairness :
Christophe Giraud (Université Paris-Saclay)
Solenne Gaucher (ENSEA – CREST)
Jean-Michel Loubes (Université de Toulouse III)
Nicolas Schreuder (CNRS, Université Gustave Eiffel))

Generative modeling :
Arnak Dalalyan (ENSAE – CREST)
Andrej Risteski (Carnegie Mellon University)

High dimensional covariance estimation :
Zhao Ren (University of Pittsburgh)
Angelika Rohde (University of Freiburg)

High dimensional testing :
Subhodh Kotekal (University of Chicago)
Yuhao Wang (Tsinghua University, Shanghai Qi Zhi Institute)

Learning theory :
Jaouad Mourtada (ENSAE-CREST)
Daniil Tiapkin (Ecole Polytechnique)

Privacy :
Tom Berrett (University of Warwick)
Vianney Perchet ( ENSAE-CREST)

 

 

Robustness :
Marco Avella Medina (Colombia University)
Chao Gao (University of Chicago)
Alexey Kroshnin (Weierstrass Institute)
Guillaume Lecué (ENSAE)
Arshak Minasyan (ENSAE-CREST)
Stanislav Minsker (University of Southern California)

Round table on fairness and robustness in AI :
Boris Noumedem Djieuzem (BNP PARIBAS)
Omar Alonso Doria Arrieta (BNP PARIBAS)
Jean-Michel Loubes (Université de Toulouse III)

SDE diffusions :
Marc Hoffmann (Université Paris-Dauphine)
Mark Podolskij (University of Luxembourg)
Markus Reiß (Humboldt University of Berlin)

Uncertainty quantification :
Eduard Belitser (Free University of Amsterdam)
Pierre Bellec (Rutgers University)
Maxim Panov (Mohamed bin Zayed University of Artificial Intelligence)

SPONSORS