BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Universal compression of images: send the model?
DTSTART:20181212T100000
DTEND:20181212T120000
DTSTAMP:20260530T053632Z
UID:a89da2458544c9dfc55f16799b4ca0e22492d457defde5fe4ef4ac0b
CATEGORIES:Conferences - Seminars
DESCRIPTION:Aline ROUMY\,  INRIA-Rennes (France)\nAline Roumy received th
 e Engineering degree from Ecole Nationale Superieure de l'Electronique et 
 de ses Applications (ENSEA)\, France in 1996\, the Master degree in 1997 a
 nd the Ph.D. degree in 2000 from the University of Cergy-Pontoise\, France
 . During 2000-2001\, she was a research associate at Princeton University\
 , Princeton\, NJ. On November 2001\, she joined INRIA\, Rennes\, France as
  a research scientist. She has held visiting positions at Eurecom and UC B
 erkeley. She serves as an Associate Editor for the Annals of telecommunica
 tions and for the IEEE Transactions on Image Processing. Her current resea
 rch interests include the area of statistical signal and image processing\
 , coding theory and information theory.\n \nABSTRACT:\nUniversal compress
 ion is the problem of compressing a data source without knowledge of the s
 ource probability distribution. \nThis occurs in particular when images a
 nd videos have to be compressed. \nAt the heart of the matter are the que
 stions of modeling the data and coding the model. \nIn the first part of 
 the presentation\, the impact of the model will be discussed and I will pr
 esent new\, learned models of images and evaluate their impact on image co
 mpression  performance. In particular\, a transform and a predictor that 
 both aim at capturing spatial correlation\, will be learned (via a deep au
 toencoder and a set of deep neural networks).\nIn the second part of the p
 resentation\, I will introduce a novel source coding problem allowing mass
 ive random access to a large database of correlated sources. I will show t
 hat it is possible to extract arbitrary sources from an appropriately comp
 ressed database purely by bit extraction\, and at the same rate as if the 
 database was decoded and the requested sources were re-encoded.\nThen the 
 question of source modeling in view of compression will be discussed.
LOCATION:ELG 120 https://plan.epfl.ch/?room=ELG120
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR