Sparse atomic feature learning via gradient regularization: With applications to finding sparse representations of fMRI activity patterns

Michael J. O'Brien, Matthew S. Keegan, Tom Goldstein, Rachel Millin, James Benvenuto, Kendrick Kay, Rajan Bhattacharyya

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

We present an algorithm, Sparse Atomic Feature Learning (SAFL), that transforms noisy labeled datasets into a sparse domain by learning atomic features of the underlying signal space via gradient minimization. The sparse signal representations are highly compressed and cleaner than the original signals. We demonstrate the effectiveness of our techniques on fMRI activity patterns. We produce low-dimensional, sparse representations which achieve over 98% compression of the original signals. The transformed signals can be used to classify left-out testing data at a higher accuracy than the initial data.

Original languageEnglish (US)
Title of host publication2014 IEEE Signal Processing in Medicine and Biology Symposium, IEEE SPMB 2014 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781479981847
DOIs
StatePublished - 2014
Externally publishedYes
Event2014 IEEE Signal Processing in Medicine and Biology Symposium, IEEE SPMB 2014 - Philadelphia, United States
Duration: Dec 13 2014Dec 13 2014

Publication series

Name2014 IEEE Signal Processing in Medicine and Biology Symposium, IEEE SPMB 2014 - Proceedings

Other

Other2014 IEEE Signal Processing in Medicine and Biology Symposium, IEEE SPMB 2014
Country/TerritoryUnited States
CityPhiladelphia
Period12/13/1412/13/14

Bibliographical note

Publisher Copyright:
© 2014 IEEE.

Fingerprint

Dive into the research topics of 'Sparse atomic feature learning via gradient regularization: With applications to finding sparse representations of fMRI activity patterns'. Together they form a unique fingerprint.

Cite this