BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Probabilistic computing applications: BayesDB and stochastic digit
 al circuits - Vikash K. Mansinghka (MIT)
DTSTART:20140404T100000Z
DTEND:20140404T110000Z
UID:TALK51687@talks.cam.ac.uk
CONTACT:Zoubin Ghahramani
DESCRIPTION:This talk consists of two shorter talks on specific probabilis
 tic computing systems:\n\n1. BayesDB\, a Bayesian database table\, lets us
 ers query many of the probable implications of their tabular data as natur
 ally as SQL lets them query the data itself. With the Bayesian Query Langu
 age (BQL)\, a domain-specific probabilistic programming language for data 
 tables\, users with no statistics training can solve basic data science pr
 oblems\, such as detecting predictive relationships between variables\, in
 ferring missing values\, simulating probable observations\, and identifyin
 g statistically similar database entries. BayesDB is based on a nonparamet
 ric Bayesian machine learning technique for directly estimating the full m
 ultivariate (joint) distribution underlying high-dimensional\, heterogeneo
 usly typed data. I will illustrate BayesDB and give an overview of current
  applications to datasets from econometrics and sociology. I will also dis
 cuss the potential for using BayesDB to improve the quality of the empiric
 al reasoning performed by non-experts and begin to mitigate the shortage o
 f analysts with expertise in statistics.\n\n2. The brain interprets ambigu
 ous sensory information faster and more reliably than modern computers\, u
 sing neurons that are slower and less reliable than logic gates. But Bayes
 ian inference\, which underpins many computational models of perception an
 d cognition\, appears computationally challenging even given modern transi
 stor speeds and energy budgets. I will show how to build fast Bayesian com
 puting machines using intentionally stochastic\, digital parts\, narrowing
  this efficiency gap by multiple orders of magnitude.
LOCATION:Engineering Department\, CBL Room BE-438
END:VEVENT
END:VCALENDAR