Online Program Home
  My Program

Abstract Details

Activity Number: 425 - SPEED: Reliable Statistical Learning and Data Science
Type: Contributed
Date/Time: Tuesday, August 1, 2017 : 3:05 PM to 3:50 PM
Sponsor: Section on Statistical Learning and Data Science
Abstract #325178
Title: Expected Conditional HSIC for Testing Independence
Author(s): Chenlu Ke* and Xiangrong Yin
Companies: and University of Kentucky
Keywords: Independence testing ; Reproducing kernel Hilbert space ; kernel method
Abstract:

We propose a novel conditional Hilbert-Schmidt Independence Criterion for testing independence between two random vectors. The relation between our index and a conditional version of distance covariance measure is studied. Two empirical estimates and corresponding independence tests are developed under different scenarios. Finally, we illustrate the advantages of our method by simulations across a variety of settings and real data applications.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2017 program

 
 
Copyright © American Statistical Association