BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Learning What to Share: Feature Learning for Adversarially Robust 
 Federated Systems
DTSTART:20260331T111500
DTEND:20260331T120000
DTSTAMP:20260407T002309Z
UID:0e4a0967aa950e989e309168b5aa52fc7b04db53c93742eebf6c3dc1
CATEGORIES:Conferences - Seminars
DESCRIPTION:Leonardo F. Toso\, fourth-year Ph.D. candidate in Electrical 
 Engineering at Columbia University\, USA.\nAbstract: Federated learning e
 nables multiple agents to collaboratively train models without sharing raw
  data. However\, real-world federated systems face two significant challen
 ges:\n (i) data heterogeneity\, where agents have fundamentally different
  data distributions\, and\n (ii) adversarial behavior\, where some agents
  may be corrupted or malicious.\n\nExisting adversarially robust federated
  systems attempt to learn a single global model while defending against ad
 versarial attacks. However\, even with an infinite number of data samples\
 , heterogeneous data limit efficient agent personalization. \n\nIn this t
 alk\, we show that the curse of heterogeneity is not fundamental. The key 
 idea is that heterogeneity often lives in the final agent-specific model
 ’s layers. Therefore\, by learning a shared feature encoder across all a
 gents and allowing each agent to personalize its local-specific parameters
 \, we can significantly overcome the effects of data heterogeneity\, even 
 under adversarial attacks.\n\nBiography:  \nLeonardo F. Toso is a four
 th-year Ph.D. candidate in Electrical Engineering at Columbia University\,
  advised by Prof. James Anderson. He is a Presidential and CAIRFI (Center 
 for AI and Responsible Financial Innovation) Fellow. During Fall 2025\, he
  served as the instructor of convex optimization at Columbia under the Tea
 ching fellowship program. His research focuses on the intersection of cont
 rol theory\, machine learning\, and optimization\, with particular emphasi
 s on meta-learning\, federated learning\, representation learning\, and ad
 aptive control. His work integrates safety\, robustness\, and learning in 
 complex distributed systems. He has received the Best Paper Award at L4DC 
 2024 and the Outstanding Paper Award at CDC 2025\, and has published exten
 sively at top venues including ICLR\, TMLR\, AAAI\, CDC\, ACC\, and L4DC.
LOCATION:ME C2 405 https://plan.epfl.ch/?room==ME%20C2%20405
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR