Computation of Percentiles From Obfuscated Data
|Event Date:||February 17, 2017|
|Speaker:||Dr. Bimal Roy|
|Speaker Affiliation:||Indian Statistical Institute Kolkata, India|
|Type:||Computer Engineering Area Seminar
|Contact Name:||Professor Saurabh Bagchi
|School or Program:||Electrical and Computer Engineering
Privacy protection and data security have received a huge amount of attention these days due to the increasing need to protect various sensitive information like credit card data, medical data etc. There are various ways to protect data; here we are interested in ways, that may as well retain its statistical uses to some extent. One such way is to mask a data with additive or multiplicative noise, and to get back to certain desired parameters of the original distribution from the knowledge of the noise distribution and masked data. In this presentation we discuss the estimation of any desired percentile of a quantitative data-set, masked with additive noise. We also propose a method to choose appropriate parameters for the noise distribution. We also discuss an application of this problem in cryptography, where data are obfuscated using a binary string generated using SHA3 hash function and how the original data points can be easily retrieved if a certain key is known.
Dr. Bimal Roy has a Ph.D. from the University of Waterloo. He is a professor at the Indian Statistical Institute at Kolkata, India and also Head, R.C.Bose Centre for Cryptology & Security at the Institute. His research interests include Cryptology, Information Security, Combinatorics, and Statistics. He has received awards that include: Padmashri (civilian award from the Government of India) and the IBM Faculty Award. He has held visiting positions at University of Ottawa(Canada), INRIA(France), Chinese Academy of Science(PRC), Kyushu University(Japan), etc. He has advisory roles to the Government of India, UNESCO, Tata Consultancy Services, etc.