BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:IC Colloquium : The Role of Neurification in Building Machines tha
 t Think
DTSTART:20161107T161500
DTEND:20161107T173000
DTSTAMP:20260916T212915Z
UID:2318f824514054e7be4b521546cee7808d120ae434d0e7b265dd98ee
CATEGORIES:Conferences - Seminars
DESCRIPTION:By : Daan Wierstra - Google DeepMind\n\nAbstract :\nBuilding m
 achines that think requires us to think `out of the box'\, as most state-o
 f-the-art machine learning algorithm families -- such as Deep Learning or 
 those of the probabilistic inference persuasion -- suffer from either high
  computational cost\, brittle assumptions or from insurmountably big data 
 requirements. This prevents the possibility of tractable one-shot learning
 \, rapid adaptability of agents to changing environments\, learned plannin
 g and the development of scalable exploration strategies and intrinsic mot
 ivation. In this talk I will highlight recent research at DeepMind aimed a
 t bridging the gap between fast\, data-hungry algorithms and slow data-eff
 icient algorithms with more explicit priors. I'll first concentrate on amo
 rtised inference methods that fuse ideas from deep learning and variationa
 l inference\, and then continue to demonstrate the viability of our `neuri
 fication' program\, that is\, the development of deep-learning-style train
 able alternatives to many canonical machine learning algorithms.\n\nBio :\
 nDaan Wierstra leads the `Frontiers' research team at Google DeepMind\, fo
 cusing efforts on deep generative models\, one-shot learning\, tractable m
 easures of uncertainty and deep memory architectures. He did his PhD with 
 Juergen Schmidhuber at IDSIA\, the Swiss AI lab in Lugano\, and his postdo
 c with Wulfram Gerstner at EPFL\, Lausanne.\n\nMore information
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR