BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Generalisation for Adaptive Data Analysis - Thomas Steinke (IBM Al
 maden Research Center)
DTSTART:20161122T153000Z
DTEND:20161122T163000Z
UID:TALK69264@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:Adaptivity is an important aspect of data analysis -- that is\
 , the choice of questions to ask about a dataset is often informed by prev
 ious use of the same dataset. However\, statistical validity is typically 
 only guaranteed in a non-adaptive model\, in which the questions must be s
 pecified before the dataset is collected. A recent line of work initiated 
 by Dwork et al. (STOC 2015) provides a formal model for studying the power
  of adaptive data analysis.  &nbsp\;  This talk will show that there are s
 ophisticated techniques -- using tools from information theory and differe
 ntial privacy -- that enable us to ensure that adaptive data analysis prov
 ides statistically valid answers that generalise to the overall population
  from which the dataset was drawn. This talk will also discuss how adaptiv
 e data analysis is inherently more powerful than non-adaptive data analysi
 s\, namely there is an exponential separation between the number of adapti
 ve queries needed to overfit a dataset and the number of non-adaptive quer
 ies needed.  &nbsp\;
LOCATION:Seminar Room 2\, Newton Institute
END:VEVENT
END:VCALENDAR