A Flexible Non-parametric Framework for High-Dimensional Multi-way Data

• Activity Number: 600 - High-Dimensional Time Series and Applications in Social and Biological Sciences
• Type: Invited
• Date/Time: Thursday, August 3, 2017 : 8:30 AM to 10:20 AM
• Sponsor: WNAR
• Author(s): Zhaoxia Yu* and Tong Shen and Dustin Pluta and Hernando Ombao
• Companies: University of California, Irvine and University of California, Irvine and University of California, Irvine and University of California, Irvine; King Abdullah University of Science and Technology
• Keywords: imaging ; genetics ; brain ; connectivity ; high-dimensional ; non-parametric

Abstract

Data collected in many scientific areas are inherently high-dimensional and multi-way. While such data provides an excellent opportunity for us to conduct an integrative analysis of multiple data modalities, it is challenging to model associations between sets of massive, complexly structured, and high-dimensional data. Here we propose a flexible and easy-to-implement non-parametric framework to assessing the overall association between high-dimensional modalities, such as between genetics and brain connectivity. The principles that we propose are applicable to various types of high dimensional data. We will also illustrate how the methods are connected to classical regression-based methods.