BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//6.4.7.2//EN
TZID:Europe/Paris
X-WR-TIMEZONE:Europe/Paris
BEGIN:VEVENT
UID:0-920@lptms.universite-paris-saclay.fr
DTSTART;TZID=Europe/Paris:20231212T110000
DTEND;TZID=Europe/Paris:20231212T120000
DTSTAMP:20231206T095755Z
URL:http://www.lptms.universite-paris-saclay.fr/seminars/seminaire-du-lptm
 s-jorge-cossio-ens/
SUMMARY:Séminaire du LPTMS : Jorge Fernandez-de-Cossio-Diaz (ENS Paris) - 
 Salle des séminaires du FAST et du LPTMS\, bâtiment Pascal n°530 - 12 D
 éc 23 11:00
DESCRIPTION:Disentangling Representations in Restricted Boltzmann Machines 
 without AdversariesJorge Fernandez-de-Cossio-Diaz (ENS Paris)A goal of uns
 upervised machine learning is to build representations of complex high-dim
 ensional data\, with simple relations to their properties. Such disentangl
 ed representations make it easier to interpret the significant latent fact
 ors of variation in the data\, as well as to generate new data with desira
 ble features. The methods for disentangling representations often rely on 
 an adversarial scheme\, in which representations are tuned to avoid discri
 minators from being able to reconstruct information about the data propert
 ies (labels). Unfortunately\, adversarial training is generally difficult 
 to implement in practice. Here we propose a simple\, effective way of dise
 ntangling representations without any need to train adversarial discrimina
 tors and apply our approach to Restricted Boltzmann Machines\, one of the 
 simplest representation-based generative models. Our approach relies on th
 e introduction of adequate constraints on the weights during training\, wh
 ich allows us to concentrate information about labels on a small subset of
  latent variables. The effectiveness of the approach is illustrated with f
 our examples: the CelebA dataset of facial images\, the two-dimensional Is
 ing model\, the MNIST dataset of handwritten digits\, and the taxonomy of 
 protein families. In addition\, we show how our framework allows for analy
 tically computing the cost\, in terms of the log-likelihood of the data\, 
 associated with the disentanglement of their representations.\n\nPaper lin
 k: https://journals.aps.org/prx/abstract/10.1103/PhysRevX.13.021003\n\n---
 --- This seminar will be on zoom also: https://universite-paris-saclay-fr.
 zoom.us/j/93026726739?pwd=TFhIdHMrSy8vRkNMdzZMT3ptSFRxQT09\nMeeting ID: 93
 0 2672 6739 Passcode: 244859\n\n
CATEGORIES:seminars
LOCATION:Salle des séminaires du FAST et du LPTMS\, bâtiment Pascal n°53
 0\, rue André Riviere\, Orsay\, 91405\, France
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=rue André Riviere\, Orsay\
 , 91405\, France;X-APPLE-RADIUS=100;X-TITLE=Salle des séminaires du FAST 
 et du LPTMS\, bâtiment Pascal n°530:geo:0,0
END:VEVENT
BEGIN:VTIMEZONE
TZID:Europe/Paris
X-LIC-LOCATION:Europe/Paris
BEGIN:STANDARD
DTSTART:20231029T020000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
END:STANDARD
END:VTIMEZONE
END:VCALENDAR