TY - GEN
T1 - Privacy-preserving source separation for distributed data using independent component analysis
AU - Imtiaz, Hafiz
AU - Silva, Rogers
AU - Baker, Bradley
AU - Plis, Sergey M.
AU - Sarwate, Anand D.
AU - Calhoun, Vince
N1 - Publisher Copyright:
© 2016 IEEE.
Copyright:
Copyright 2017 Elsevier B.V., All rights reserved.
PY - 2016/4/26
Y1 - 2016/4/26
N2 - Building good feature representations and learning hidden source models typically requires large sample sizes. In many applications, however, the size of the sample at an individual data holder may not be sufficient. One such application is neuroimaging analyses for mental health disorders - there are many individual research groups, each with a moderate number of subjects. Pooling such data can enable efficient feature learning, but privacy concerns prevent sharing the underlying data. We propose a model for private feature learning in which the data holders share differentially private views of their respective datasets to enable collaborative learning of a joint feature map. We give an example of such an algorithm for independent component analysis (ICA) - a popular blind source separation algorithm used in neuroimaging analyses. Our algorithm is a differentially private version of the recently proposed distributed joint ICA algorithm. We evaluate the performance of this method on simulated functional magnetic resonance imaging (fMRI) data.
AB - Building good feature representations and learning hidden source models typically requires large sample sizes. In many applications, however, the size of the sample at an individual data holder may not be sufficient. One such application is neuroimaging analyses for mental health disorders - there are many individual research groups, each with a moderate number of subjects. Pooling such data can enable efficient feature learning, but privacy concerns prevent sharing the underlying data. We propose a model for private feature learning in which the data holders share differentially private views of their respective datasets to enable collaborative learning of a joint feature map. We give an example of such an algorithm for independent component analysis (ICA) - a popular blind source separation algorithm used in neuroimaging analyses. Our algorithm is a differentially private version of the recently proposed distributed joint ICA algorithm. We evaluate the performance of this method on simulated functional magnetic resonance imaging (fMRI) data.
UR - http://www.scopus.com/inward/record.url?scp=84992406261&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84992406261&partnerID=8YFLogxK
U2 - 10.1109/CISS.2016.7460488
DO - 10.1109/CISS.2016.7460488
M3 - Conference contribution
AN - SCOPUS:84992406261
T3 - 2016 50th Annual Conference on Information Systems and Sciences, CISS 2016
SP - 123
EP - 127
BT - 2016 50th Annual Conference on Information Systems and Sciences, CISS 2016
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 50th Annual Conference on Information Systems and Sciences, CISS 2016
Y2 - 16 March 2016 through 18 March 2016
ER -