BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//RLASKEY//CALENDEROUS//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VEVENT
DTSTAMP:20260814T104729Z
LAST-MODIFIED:20140422T151814Z
DTSTART:20140423T143000Z
DTEND:20140423T153000Z
UID:event1263@bu.edu
URL:http://physics.bu.edu/internal/events/show/1263
SUMMARY:CMT Group Meeting
DESCRIPTION:Featuring Pankaj Mehta\, Boston University\n\nPart of the Conde
	nsed Matter Theory Seminar Series.\n\nDeep Learning as Renormalization and 
	Holography\n\nDeep Learning is one of the most promising Machine Learning a
	nd Artificial Intelligence (AI) techniques for extracting important feature
	s from large data sets (  http://www.nytimes.com/2012/11/24/science/scienti
	sts-see-advances-in-deep-learning-a-part-of-artificial-intelligence.html?sm
	id=pl-share&_r=2&  ;   http://www.technologyreview.com/news/524026/is-googl
	e-cornering-the-market-on-deep-learning/). Deep Learning employs a neural-i
	nspired architecture with multiple layers and are often trained using an it
	erative\, layer-by-layer algorithm. Despite their tremendous success\, the 
	basic logic behind Deep Architectures is not well understood. Here\, we sho
	w that there is an exact mapping between Deep Learning and real space Renor
	malization Group techniques from physics. Furthermore\, we show that the re
	sulting Deep Architectures have a natural interpretation as  holographic du
	als. We use this correspondence to give a general proof of the AdS-CFT corr
	espondence. We illustrate these ideas using the 1D and 2D Ising Models.
LOCATION:SCI 328\, 590 Commonwealth Avenue\, 02215
STATUS:CONFIRMED
CLASS:PUBLIC
END:VEVENT
END:VCALENDAR