BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Machine Learning @ CUED
SUMMARY:Application of the optimal transport Gromov-Wasser
 stein problem to manifold learning and graph analy
 sis - Hugues Van Assel\; École Normale Supérieure 
 de Lyon
DTSTART;TZID=Europe/London:20240613T110000
DTEND;TZID=Europe/London:20240613T120000
UID:TALK217963AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/217963
DESCRIPTION:I will present some of my recent PhD research\, wh
 ere we reexamine unsupervised learning methods thr
 ough the perspective of distributions using optima
 l transport. By drawing connections with the Gromo
 v-Wasserstein (GW) problem\, this research introdu
 ces a new comprehensive framework called distribut
 ional reduction. This framework encompasses both d
 imensionality reduction (DR) and clustering as spe
 cial cases\, enabling to address them jointly with
 in a single optimization problem. Additionally\, I
  will discuss the applications of the GW problem a
 s a similarity measure between structured data rep
 resented as distributions\, typically lying in dif
 ferent metric spaces\, such as graphs of varying s
 izes. This will naturally lead to exploring resear
 ch directions in graph generative modeling using G
 W.\n\n\nShort biography:\n\nI studied at École Pol
 ytechnique and hold a research master's degree fro
 m École Normale Supérieure of Paris-Saclay. Curren
 tly\, I am a third-year PhD student in the mathema
 tics department at École Normale Supérieure de Lyo
 n\, supervised by Aurélien Garivier and Titouan Va
 yer. I have completed several internships\, includ
 ing positions at IBM Research Paris\, where I work
 ed on time series forecasting\; Huawei Noah’s Ark 
 Lab London\, focusing on model-based reinforcement
  learning with normalizing flows\; and Institut Pa
 steur\, where I researched variational autoencoder
 s (VAEs) for biosignals.
LOCATION:Cambridge University Engineering Department\, CBL 
 Seminar room BE4-38.
CONTACT:Dr H Ge
END:VEVENT
END:VCALENDAR