BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Privacy-preserving algorithms for signal processing and machine le
 arning
DTSTART:20131024T161500
DTEND:20131024T171500
DTSTAMP:20260508T010627Z
UID:a59469388f0c4331150b92f532703b8db20e0f3b75a3ac4b23d34a36
CATEGORIES:Conferences - Seminars
DESCRIPTION:Prof. Anand Sarwate\, Toyota\nBio: Anand Sarwate is currently 
 a Research Assistant Professor at the Toyota Technological Institute at Ch
 icago\, a philanthropically endowed academic institute located on the Univ
 ersity of Chicago campus. Prior to that he was a postdoc in the Informatio
 n Theory and Applications Center (ITA) at UC San Diego.  He received his 
 PhD from UC Berkeley in 2008\, and undergraduate degrees in Mathematics an
 d Electrical Engineering from MIT in 2002.  He received the Demetri Angel
 akos and Samuel Silver awards from the EECS department at UC Berkeley.  H
 e is broadly interested in algorithms applied to problems in distributed s
 ystems\, signal processing\, machine learning\, statistics\, and privacy a
 nd security.  He will be joining as an Assistant Professor in the Departm
 ent of Electrical and Computer Engineering in January 2014.\nLarge-scale d
 ata mining of private information (such as medical or financial records) h
 as the potential to revolutionize the way we live. On the positive side\, 
 it can enable large-scale studies on the comparative effectiveness of trea
 tments\, novel risk factors for diseases\, and patient-centered personaliz
 ed medicine.  There are many challenges to overcome in order to realize t
 hese benefits\, ranging from the way we represent data to legal and policy
  arrangements between medical institutions.  Uniting these is a concern a
 bout the privacy -- we want to guarantee low privacy risk and high precisi
 on\, or utility.  The fundamental question to answer is this : how much d
 ata do we need to guarantee acceptable levels privacy and utility?\nIn thi
 s talk I describe practical approaches for managing this tradeoff.  These
  methods guarantee differential privacy\, a cryptographically-motivated de
 finition of privacy which has been widely adopted in the computer science 
 community.  I will discuss basic ideas from differential privacy and how 
 to use them to build algorithms for classification and dimensionality redu
 ction\, which are two of the most common tasks in machine learning.  I wi
 ll also describe some exciting future prospects for privacy-preserving alg
 orithms in signal processing\, optimization\, and learning.
LOCATION:INR113 http://plan.epfl.ch/?zoom=20&recenter_y=5863814.94355&rece
 nter_x=730548.85489&layerNodes=fonds\,batiments\,labels\,information\,park
 ings_publics\,arrets_metro&floor=1&q=INR113
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR