BEGIN:VCALENDAR
VERSION:2.0
CALSCALE:GREGORIAN
PRODID:iCalendar-Ruby
BEGIN:VEVENT
CATEGORIES:Colloquium/Seminar
DESCRIPTION:Randomization methods for big-data\n\nAbstract: In this era of
big-data\, we must adapt our algorithms to handle large datasets. One obvio
us issue is that the number of floating-point operations (flops) increases
as the input size increases\, but there are many less obvious issues as wel
l\, such as the increased communication cost of moving data between differe
nt levels of computer memory. Randomization is increasingly being used to a
lleviate some of these issues\, as those familiar with random mini-batch sa
mpling in machine learning are well aware of. This talk goes into some spec
ific examples of using randomization to improve algorithms. We focus on spe
cial classes of structured random dimensionality reduction\, including the
count sketch\, tensorSketch\, Kronecker fast Johnson-Lindenstrauss sketch\,
and pre-conditioned sampling. These randomized techniques can then be appl
ied to speeding up the classical Lloyd's algorithm for K-means and for comp
uting tensor decompositions\, for example. If time permits\, we will also s
how extensions to optimization\, including a gradient-free method that uses
random finite differences and a method for solving semi-definite programs
in an optimal low-memory fashion.\n\nBio: Stephen Becker is an associate pr
ofessor of applied mathematics at the University of Colorado Boulder\, with
courtesy appointments in the Electrical\, Computer and Energy Engineering
and Computer Science departments. Previously he was a Herman Goldstine Post
doctoral fellow in Mathematical Sciences at IBM Research in Yorktown Height
s\, NY\, and a postdoctoral fellow via the Fondation Sciences Mathématiques
de Paris at Paris 6. He received his PhD in 2011 from Caltech under Emmanu
el Candès\, and bachelor’s degrees in math and physics from Wesleyan Univer
sity. His research interests are in optimization\, machine learning\, signa
l processing\, imaging\, inverse problems in quantum information\, PDE-cons
trained optimization\, and randomized numerical linear algebra.\n\nhttps://
cuboulder.zoom.us/j/190280621
DTEND:20231026T223000Z
DTSTAMP:20240912T043453Z
DTSTART:20231026T213000Z
GEO:40.006791;-105.262818
LOCATION:Engineering Center\, ECCR 265
SEQUENCE:0
SUMMARY:CS Colloquium: Stephen Becker on Randomization methods for big-data
UID:tag:localist.com\,2008:EventInstance_44456031506658
URL:https://calendar.colorado.edu/event/cs_colloquium_stephen_becker_on_ran
domization_methods_for_big-data
END:VEVENT
END:VCALENDAR