BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Workshop: Generative Neural Networks with Applications to Imaging
DTSTART:20240223T130000
DTEND:20240223T150000
DTSTAMP:20260415T230420Z
UID:864ab8f431a7cec888c1a651299b9297568091acdd2ec2237147266c
CATEGORIES:Conferences - Seminars
DESCRIPTION:Prof. Dr. Ullrich Köthe\, Heidelberg University \n<< THE EV
 ENT IS SOLD OUT >>\n\nThe workshop is jointly organised by the EPFL Center
  for Imaging and the EPFL AI Center. \n\nAbstract:\nDeep generative mo
 dels have emerged as a powerful paradigm for hard problems in computer vis
 ion\, natural language processing and the sciences. "Generative" here refe
 rs to two capabilities: (1) generate synthetic data that are (ideally) ind
 istinguishable from real data\, and (2) calculate the probability density 
 of any given data point. These models thus acquire a high degree of unders
 tanding of the phenomenon under study\, which can be exploited for novel s
 olutions to high-dimensional probability modeling\, Bayesian inference\, a
 nd interpretability\, to name just a few.\n\nThe first part of the talk wi
 ll report about leading approaches to generative modeling (normalizing flo
 ws\, rectangular flows\, and diffusion models)\, introduce their theoretic
 al and algorithmic foundations\, and describe their most successful realiz
 ations by neural networks. It will be shown that a categorization in terms
  of change-of-variables formulas associated with these methods allows for 
 a systematic understanding of many differences and similarities.\n\nThe se
 cond part will be devoted to applications of generative models as a tool f
 or other machine learning tasks\, e.g. generative image classification and
  the solution of inverse problems. Most notably\, generative models allow 
 for a fully Bayesian treatment of these tasks and therefore lead to trustw
 orthy uncertainty quantification of the solutions. Examples from a wide ra
 nge of applications\, from medical imaging to psychology\, will illustrate
  the power of the resulting framework.\n\nBio:\nUllrich Köthe received th
 e Diploma degree in physics from the University of Rostock\, Rostock\, Ger
 many\, in 1991\, and the Ph.D. degree in computer science from the Univers
 ity of Hamburg\, Hamburg\, Germany\, in 2000.\,He is currently an Adjunct 
 Professor of computer science with the Interdisciplinary Center for Scient
 ific Computing\, Heidelberg University\, Heidelberg\, Germany. His researc
 h focuses on the connection between machine learning and the sciences from
  a methodological perspective and an application perspective and\, in part
 icular\, on the interpretability of machine learning results.\n\nRegistrat
 ion required
LOCATION:BC 01 https://plan.epfl.ch/?room==BC%2001
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR