Title: Joint and individual non-Gaussian component analysis
Authors: Benjamin Risk - Emory University (United States) [presenting]
Irina Gaynanova - Texas A and M University (United States)
Abstract: As advances in technology allow the acquisition of complementary information, it is increasingly common for scientific studies to collect multiple data sets. Large-scale neuroimaging studies often include multiple imaging modalities (e.g., functional MRI, diffusion MRI, and/or structural MRI) and behavioral data, with the aim to understand the relationships between data sets. Common approaches to data integration utilize transformations that maximize covariance or correlation, but measures of information using higher order moments may reveal additional structure. We introduce Joint and Individual Non-Gaussian component analysis (JIN) for data integration. We focus on information shared in subject score subspaces estimated using non-quadratic nonlinearities, and we also examine information unique to each data set. We apply our method to data from the Human Connectome Project.