BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Advanced machine learning algorithms for high-dimensional remote s
 ensing image processing
DTSTART:20100426T161500
DTSTAMP:20260407T020824Z
UID:ab7ebb0976fb06328aca4e7bdcaef13f8c12da4d59e1da4c146d8342
CATEGORIES:Conferences - Seminars
DESCRIPTION:Tuia\, Devis (Université de Lausanne-UNIL)\nThe technical dev
 elopments in recent years have brought the quantity and quality of digital
  information to an unprecedented level\, as enormous archives of satellite
  images are available to the users. However\, even if these advances open 
 more and more possibilities in the use of digital imagery\, they also rise
  several problems of storage and treatment. The latter is considered in th
 is presentation: the processing of very high spatial and spectral resoluti
 on images is treated with approaches based on data-driven algorithms relyi
 ng on kernel methods.\nIn particular\, the problem of image classification
 \, i.e. the categorization of the image's pixels into a reduced number of 
 classes reflecting spectral and contextual properties\, is studied through
  the different models presented. The accent is put on algorithmic efficien
 cy and the simplicity of the approaches proposed\, to avoid too complex mo
 dels that would not be used by users.\nAfter a short introduction on machi
 ne learning for remote sensing\, two problems will be considered : first\,
  the question of high dimensionality and collinearity of the image feature
 s is studied by an adaptive model learning the relevant image features. Th
 is model provides automatically an accurate classifier and a ranking of th
 e relevance of the single features. Second is the question of scarcity and
  unreliability of the labeled information: when confronted to such situati
 ons\, the user can either construct the labeled set iteratively by direct 
 interaction with the machine (active learning) or use the unlabeled data t
 o increase robustness and quality of the description of data (semi-supervi
 sed learning). Both solutions will be discussed during the presentation an
 d advantages and limitations of the different strategies will be pointed o
 ut.
LOCATION:GR B3 30
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR