Activity Number:
|
623
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, August 3, 2016 : 2:00 PM to 3:50 PM
|
Sponsor:
|
Section on Medical Devices and Diagnostics
|
Abstract #320670
|
|
Title:
|
Hierarchical Classification Models for Functional Data
|
Author(s):
|
Abhirup Mallik* and Snigdhansu Chatterjee
|
Companies:
|
and University of Minnesota - Twin Cities
|
Keywords:
|
Functional Data ;
Bayesian ;
EEG
|
Abstract:
|
In many modern applications data is collected continuously from a process over a period of time on a fine grid. Such datasets can be conceptualized to be made up of a collection of continuous functions and are known as functional data. In the field of medical imaging, functional data is extremely common and the problem of classification arises naturally. However because of the experimental setup, such datasets often have special structures based on the treatments and effects of covariates. We propose a way of modeling these data based on a hierarchical Bayesian framework that would be flexible enough to handle such structures. We also explore robust estimators of covariance matrix in these context. We discuss the proposed method for the problem of activity recognition and analysis of EEG datasets.
|
Authors who are presenting talks have a * after their name.