Brain Imaging Analysis with Sparse Models
Irina Rish, Ph.D.
IBM T.J. Watson Research Center Computational Biology Center
Yorktown Heights, NY 10598
Sparse models combine variable selection with parameter estimation when learning from from data. In small-sample high-dimensional problems, this leads to improved generalization accuracy combined with better interpretability of a model. In this talk, I will summarize our recent work on sparse models, including sparse regression and sparse Gaussian Markov Random Fields (GMRF), in neuroimaging applications, such as functional MRI data analysis, where the central objective is to gain a better insight into brain functioning, besides just learning “black-box” predictive models of mental states from imaging data.