BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Privacy-preserving Machine Learning
DTSTART:20190821T140000
DTEND:20190821T160000
DTSTAMP:20260407T110324Z
UID:ed7bde51b502e891d93e7b05a09ea45d80567a65f9714f71d5a3476a
CATEGORIES:Conferences - Seminars
DESCRIPTION:Valentin Hartmann\nEDIC candidacy exam\nExam president: Prof. 
 Carmela Troncoso\nThesis advisor: Prof. Robert West\nCo-examiner: Prof. Ma
 rtin Jaggi\n\nAbstract\nIn a world where machine learning (ML) with its ne
 ed for huge training datasets has become ubiquitous\, massive data collect
 ion has become the norm rather than the exception. With new data breaches 
 being reported on almost daily\, and details on the usage of data by compa
 nies surfacing\, the problem of ensuring privacy is not anymore of interes
 t only for researchers\, but also gets the attention of the general public
 .\n\nIt is time for a paradigm shift: Nowadays' ML methods need to be repl
 aced with privacy-preserving ones. In this proposal\, we present the first
  steps that have been made in this direction: (1) the definition of differ
 ential privacy (DP)\, a both strong and practical quantification of privac
 y\; (2) a method for training neural networks with DP\; and (3) a method f
 or creating synthetic datasets with DP from sensitive data that are almost
  as well suited for ML tasks as the original data. We conclude by showing 
 concrete directions for extending these methods to make them applicable to
  a broader class of ML tasks.\n\nBackground papers\nCalibrating noise to s
 ensitivity in private data analysis\, by Dwork\, Cynthia\, et al. Theory o
 f cryptography conference. Springer\, Berlin\, Heidelberg\, 2006.\nDeep le
 arning with differential privacy\, by  Abadi\, Martin\, et al. Proceedings
  of the 2016 ACM SIGSAC Conference on Computer and Communications Security
 . ACM\, 2016.\nPlausible deniability for privacy-preserving data synthesis
 \, by Bindschaedler\, Vincent\, Reza Shokri\, and Carl A. Gunter. Proceedi
 ngs of the VLDB Endowment 10.5 (2017): 481-492.\n\n 
LOCATION:INN 326 https://plan.epfl.ch/?room==INN%20326
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR