BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//7.2.3.1//EN
TZID:Europe/Paris
X-WR-TIMEZONE:Europe/Paris
BEGIN:VEVENT
UID:3148@i2m.univ-amu.fr
DTSTART;TZID=Europe/Paris:20191108T140000
DTEND;TZID=Europe/Paris:20191108T150000
DTSTAMP:20191024T120000Z
URL:https://www.i2m.univ-amu.fr/evenements/juho-rousu-sparse-non-linear-ca
 nonical-correlation-analysis/
SUMMARY: (...): Juho ROUSU - Sparse non-linear canonical correlation analys
 is
DESCRIPTION:: Juho ROUSU (Aalto University\, Helsinki)Canonical correlation
  analysis (CCA) methods find multivariate relations in two-view data setti
 ngs. CCA can be seen as a relative of principal component analysis\, when 
 the objective is to explain covariance between two views rather than varia
 nce within one view. When applying CCA methods\, the interest is typically
  to find the related variables in the two views\, and to uncover the relat
 ion that couples the variables. In general\, it is difficult to achieve th
 ese objectives if the underlying multivariate relation is non-linear and t
 he data is high-dimensional. CCA methods in the literature tend to be eith
 er non-linear (such as kernel CCA and deep CCA) or sparse (sparse CCA) but
  not both. In this presentation\, I discuss recent progress in learning CC
 A models where the underlying relations are both non-linear and sparse. Ou
 r new gradKCCA method is based on mapping a sparse projection of the data 
 through a non-linear kernel function. From an another angle\, we can view 
 the method as solving an ‘optimal preimage’ problem for KCCA. The mode
 l is optimized using an alternating projected gradient algorithm. Unlike K
 CCA\, the method does not rely on a kernel matrix\, and hence scales up ef
 fortless to large datasets. Experimentally\, gradKCCA shows competitive pr
 edictive performance to KCCA and deep CCA models\, while being orders of m
 agnitude faster to optimize.Reference:Uurtio\, V.\, Bhadra\, S. and Rousu\
 , J.\, 2019\, Large-scale sparse kernel canonical correlation analysis\, I
 CML’2019BioJuho Rousu is a Professor of Computer Science at Aalto Univer
 sity\, Finland. Rousu obtained his PhD in 2001 form University of Helsinki
 \, while working at VTT Technical Centre of Finland. In 2003-2005 he was a
  Marie Curie Fellow at Royal Holloway University of London. In 2005-2011 h
 e held Lecturer and Professor positions at University of Helsinki\, before
  moving to Aalto University in 2012 where he leads a research group on Ker
 nel Methods\, Pattern Analysis and Computational Metabolomics (KEPACO). Ro
 usu’s main research interest is in learning with multiple and structured
  targets\, multiple views and ensembles\, with methodological emphasis in 
 regularised learning\, kernels and sparsity\, as well as efficient convex/
 non-convex optimisation methods. His applications of interest include meta
 bolomics\, biomedicine\, pharmacology and synthetic biology.https://people
 .aalto.fi/juho.rousu
END:VEVENT
BEGIN:VTIMEZONE
TZID:Europe/Paris
X-LIC-LOCATION:Europe/Paris
BEGIN:STANDARD
DTSTART:20191027T020000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
END:STANDARD
END:VTIMEZONE
END:VCALENDAR