BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//jEvents 2.0 for Joomla//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:STANDARD
DTSTART:20230118T140000
RDATE:20230312T030000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:America/New_York EST
END:STANDARD
BEGIN:STANDARD
DTSTART:20231105T010000
RDATE:20240310T030000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:America/New_York EST
END:STANDARD
BEGIN:STANDARD
DTSTART:20241103T010000
RDATE:20250309T030000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:America/New_York EST
END:STANDARD
BEGIN:STANDARD
DTSTART:20251102T010000
RDATE:20260308T030000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:America/New_York EST
END:STANDARD
BEGIN:STANDARD
DTSTART:20261101T010000
RDATE:20270314T030000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:America/New_York EST
END:STANDARD
BEGIN:STANDARD
DTSTART:20271107T010000
RDATE:20280312T030000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:America/New_York EST
END:STANDARD
BEGIN:STANDARD
DTSTART:20281105T010000
RDATE:20290311T030000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:America/New_York EST
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20230312T030000
RDATE:20231105T010000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:America/New_York EDT
END:DAYLIGHT
BEGIN:DAYLIGHT
DTSTART:20240310T030000
RDATE:20241103T010000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:America/New_York EDT
END:DAYLIGHT
BEGIN:DAYLIGHT
DTSTART:20250309T030000
RDATE:20251102T010000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:America/New_York EDT
END:DAYLIGHT
BEGIN:DAYLIGHT
DTSTART:20260308T030000
RDATE:20261101T010000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:America/New_York EDT
END:DAYLIGHT
BEGIN:DAYLIGHT
DTSTART:20270314T030000
RDATE:20271107T010000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:America/New_York EDT
END:DAYLIGHT
BEGIN:DAYLIGHT
DTSTART:20280312T030000
RDATE:20281105T010000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:America/New_York EDT
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:18d46679e5096bb088209f64eb95421b
CATEGORIES:Special Colloquium
CREATED:20240108T153247
SUMMARY:Deep Learning Meets Sparse Regularization
LOCATION:Hill 705
DESCRIPTION:<p><span style="color: #242424; font-family: 'Segoe UI', 'Segoe UI Web (Wes
 t European)', 'Segoe UI', -apple-system, BlinkMacSystemFont, Roboto, 'Helve
 tica Neue', sans-serif; font-size: 14.6667px; font-style: normal; font-weig
 ht: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent
 : 0px; text-transform: none; widows: 2; word-spacing: 0px; white-space: nor
 mal; background-color: #ffffff; float: none;">Abstract:</span><br aria-hidd
 en="true" style="color: #242424; font-family: 'Segoe UI', 'Segoe UI Web (We
 st European)', 'Segoe UI', -apple-system, BlinkMacSystemFont, Roboto, 'Helv
 etica Neue', sans-serif; font-size: 14.6667px; font-style: normal; font-wei
 ght: 400; letter-spacing: normal; orphans: 2; text-align: start; text-inden
 t: 0px; text-transform: none; widows: 2; word-spacing: 0px; white-space: no
 rmal; background-color: #ffffff;"><br aria-hidden="true" style="color: #242
 424; font-family: 'Segoe UI', 'Segoe UI Web (West European)', 'Segoe UI', -
 apple-system, BlinkMacSystemFont, Roboto, 'Helvetica Neue', sans-serif; fon
 t-size: 14.6667px; font-style: normal; font-weight: 400; letter-spacing: no
 rmal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none
 ; widows: 2; word-spacing: 0px; white-space: normal; background-color: #fff
 fff;"><span style="color: #242424; font-family: 'Segoe UI', 'Segoe UI Web (
 West European)', 'Segoe UI', -apple-system, BlinkMacSystemFont, Roboto, 'He
 lvetica Neue', sans-serif; font-size: 14.6667px; font-style: normal; font-w
 eight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-ind
 ent: 0px; text-transform: none; widows: 2; word-spacing: 0px; white-space: 
 normal; background-color: #ffffff; float: none;">Deep learning has been wil
 dly successful in practice and most state-of-the-art artificial intelligenc
 e systems are based on neural networks. Lacking, however, is a rigorous mat
 hematical theory that adequately explains the amazing performance of deep n
 eural networks. In this talk, I present a new mathematical framework that p
 rovides the beginning of a deeper understanding of deep learning. This fram
 ework precisely characterizes the functional properties of trained neural n
 etworks. The key mathematical tools which support this framework include tr
 ansform-domain sparse regularization, the Radon transform of computed tomog
 raphy, and nonlinear approximation theory, which are all deeply rooted in h
 armonic analysis. This framework explains the effect of weight decay regula
 rization in neural network training, the importance of skip connections and
  low-rank weight matrices in network architectures, the role of sparsity in
  neural networks, and explains why neural networks can perform well in high
 -dimensional problems.</span></p>
CONTACT:Rahul Parhi, EPFL
DTSTAMP:20260826T061009
DTSTART;TZID=America/New_York:20240119T140000
DTEND;TZID=America/New_York:20240119T150000
SEQUENCE:0
TRANSP:OPAQUE
END:VEVENT
END:VCALENDAR