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:2662@i2m.univ-amu.fr
DTSTART;TZID=Europe/Paris:20190111T140000
DTEND;TZID=Europe/Paris:20190111T150000
DTSTAMP:20181227T130000Z
URL:https://www.i2m.univ-amu.fr/evenements/combining-complex-wavelets-with
 -deep-networks-aiming-to-improve-learning-efficiency-for-vision-systems/
SUMMARY: (...): Combining Complex Wavelets with Deep Networks: aiming to im
 prove learning efficiency for vision systems
DESCRIPTION:: Scattering networks [Bruna & Mallat\, IEEE Trans PAMI 2013\; 
 Oyallon & Mallat\, CVPR 2015] may be interpreted as convolutional network 
 layers in which the filters are defined by complex wavelet transforms and 
 whose layer non-linearities are typically complex modulus (L2-norm) operat
 ors. Usually they are pre-designed using standard complex wavelet design m
 ethodologies that are based on accumulated human knowledge about vision sy
 stems\, and they involve minimal training. It is found that several layers
  of scatternet can usefully replace the early layers of a deep convolution
  neural net (CNN). The aim of this strategy is that the deterministic and 
 complete nature of the wavelet transformations will result in deep network
 s that are faster at learning\, more comprehensible in their behaviour and
  perhaps better at generalisation than a CNN which has to learn all of its
  layers from finite amounts of training data. Furthermore\, by employing t
 ight-frame overcomplete wavelets and L2-norm nonlinearities\, signal energ
 y may be conserved through the scatternet layers\, leading to some interes
 ting strategies for subspace selection.-In this talk we shall suggest a nu
 mber of ways that dual-tree complex wavelets may be incorporated into deep
  networks\, either to generate scatternet front-ends or to produce interes
 ting alternatives to standard convolutional layers\, embedded deeper in th
 e network. We will also show how recent ideas on CNN layer visualisation c
 an be extended to include the wavelet-based layers too. We shall pose more
  questions than answers\, while also presenting a few results from current
  stages of this work. I am very grateful to my co-researchers on this proj
 ect\, Amarjot Singh and Fergal Cotter.http://sigproc.eng.cam.ac.uk/Main/NG
 Khttp://scholar.google.co.uk/citations?user=31Xh_68AAAAJ&hl=en
END:VEVENT
BEGIN:VTIMEZONE
TZID:Europe/Paris
X-LIC-LOCATION:Europe/Paris
BEGIN:STANDARD
DTSTART:20181028T020000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
END:STANDARD
END:VTIMEZONE
END:VCALENDAR