File(s) under permanent embargo
Differentially private multi-task learning
conference contribution
posted on 2016-01-01, 00:00 authored by Santu RanaSantu Rana, Sunil GuptaSunil Gupta, Svetha VenkateshSvetha VenkateshPrivacy restrictions of sensitive data repositories imply that the data analysis is performed in isolation at each data source. A prime example is the isolated nature of building prognosis models from hospital data and the associated challenge of dealing with small number of samples in risk classes (e.g. suicide) while doing so. Pooling knowledge from other hospitals, through multi-task learning, can alleviate this problem. However, if knowledge is to be shared unrestricted, privacy is breached. Addressing this, we propose a novel multi-task learning method that preserves privacy of data under the strong guarantees of differential privacy. Further, we develop a novel attribute-wise noise addition scheme that significantly lifts the utility of the proposed method. We demonstrate the effectiveness of our method with a synthetic and two real datasets.
History
Event
Intelligence and Security Informatics. Pacific-Asia Workshop (11th : 2016 : Auckland, New Zealand)Source
Intelligence and security informatics : 11th Pacific Asia workshop, PAISI 2016 Auckland, New Zealand, April 19, 2016 proceedingsVolume
9650Series
Lecture notes in computer science; v.9650Pagination
101 - 113Publisher
SpringerLocation
Auckland, New ZealandPlace of publication
Berlin, GermanyPublisher DOI
Start date
2016-04-19End date
2016-04-19ISSN
0302-9743eISSN
1611-3349ISBN-13
9783319318639Language
engPublication classification
B Book chapter; E Conference publication; E1 Full written paper - refereedCopyright notice
2016, Springer International PublishingExtent
14Editor/Contributor(s)
M Chau, A Wang, H ChenTitle of proceedings
PAISI 2016 : Intelligence and Security Informatics : Proceedings of the 11th Pacific-Asia WorkshopUsage metrics
Categories
No categories selectedKeywords
Licence
Exports
RefWorks
BibTeX
Ref. manager
Endnote
DataCite
NLM
DC