Abstract

Machine learning techniques have gained prominence for the analysis ofresting-state functional Magnetic Resonance Imaging (rs-fMRI) data. Here, wepresent an overview of various unsupervised and supervised machine learningapplications to rs-fMRI. We present a methodical taxonomy of machine learningmethods in resting-state fMRI. We identify three major divisions ofunsupervised learning methods with regard to their applications to rs-fMRI,based on whether they discover principal modes of variation across space, timeor population. Next, we survey the algorithms and rs-fMRI featurerepresentations that have driven the success of supervised subject-levelpredictions. The goal is to provide a high-level overview of the burgeoningfield of rs-fMRI from the perspective of machine learning applications.

 

Introduction (beta)

None

 

Conclusion (beta)

None