profile-pic
Days on Distributed Learning
September 21th and 22th, Lyon, France

Speakers


About


The GDR RSD organizes in association with SIF two days on Distributed Learning from September 21th to September 22th, 2023 in Lyon, France. Registration fees are 100€ (50% discount for SIF members) and includes breaks, social diner, and keynotes (places are limited,please note that the registration policy is "first-in-first-in").

These days will feature 8 technical keynotes (September 21th), and 8 multidisciplinary presentations dealing with the different challenges of decentralized learning (September 22th).

Registration


These theme days are also collocated with a summer school on decentralized learning.

Program


Thursday 21th September

08:30 - 09:00 Welcome/Breakfast

Session 1: Distributed Systems

09:00 - 09:30 Sara Bouchenak (INSA-Lyon)
Towards Privacy, Robustness and Fairness in Federated Learning
09:30 - 10:00 Antoine Boutet (INSA-Lyon)
When an adversary tries to steal the meager accuracy gain offered by FL
10:00 - 10:30 Discussion
10:30 - 11:00 Break

Session 2: ML and Optimisation

11:00 - 11:30 Aymeric Dieuleveut (Ecole Polytechnique)
A Tour on Compression in Federated Learning
11:30 - 12:00 Marc Tommasi (Université de Lille)
Topology Learning for Decentralized Optimization
12:00 - 12:30 Discussion
12:30 - 14:00 Lunch

Session 3: Security and Privacy

14:00 - 14:30 Chuan XU (Inria)
Passive attacks for federated learning
14:30 - 15:00 Sara Tucci (CEA)
Byzantine resilient distribted systems
15:00 - 15:30 Discussion
15:30 - 16:00 Break

Session 4: Applications and Platforms

16:00 - 16:30 Stefan Duffner (INSA-Lyon)
Federated Learning for Health Applications
16:30 - 17:00 Kevin Scaman (Inria)
The Impact of Data Heterogeneity on Personalized Federated Learning
17:00 - 17:30 Discussion



Social Event: Brasserie Georges, 20h



Friday 22th September (9h - 14h)

08:30 - 09:00 Welcome/Breakfast

09:00 - 09:30 Christian Reimsbach-Kounatze (OCDE)
Emerging privacy-enhancing technologies - Current regulatory and policy approaches
09:30 - 10:00 Claude Kirchner (CNPEN)
Enjeux d’éthique des architectures numériques décentralisées, ou pas.
10:00 - 10:30 Margo Bernelin (CNRS)
L'apprentissage distribué et le droit : nage synchronisée ?
10:30 - 11:00 Break
11:00 - 11:30 Alexis Leautier (CNIL)
Avantages pour la vie privée des approches décentralisées et recommandations de la CNIL
11:30 - 11:45 Antoine Richard (HCL)
Problématiques de l’apprentissage fédéré en santé (le cas d’un réseau de CHU)
11:45 - 12:00 Clément Devic (Unicancer)
Retour d'expérience: le cas d'un apprentissage fédéré en imagerie TEP-TDM pour prédire la survie de patients atteints de cancer pulmonaire métastatique traité par immunothérapie
12:00 - 12:45 Table Ronde : Problématiques de l’apprentissage fédéré en santé
Margo Bernelin (CNRS), Alexis Leautier (CNIL), Claude Kirchner (CNPEN), François Talbot (HCL), Antoine Richard (HCL), Christian Reimsbach-Kounatze (OCDE), Marco Lorenzi (Inria), Clément Devic (Unicancer), Olivier Humbert (Unicancer)
13:00 - 14:00 Lunch












Organizers





Sonia Ben Mokhtar (Insa-Lyon, LIRIS)

Laurent Réveillère (Université de Bordeaux, SIF)

Antoine Boutet (Insa-Lyon, CITI)

Carole Frindel (Insa-Lyon, CREATIS)

Eddy Caron (ENS-Lyon, LIP)