Date |
Dec 18, 2017 |
Speaker |
Dr. Yingying Xu,
Postdoctoral fellow, HIIT (Helsinki Institute of Information
Technology), Finland
|
Title |
Inverse finite-size scaling for high-dimensional significance analysis
|
Abstract |
We propose an efficient procedure for significance
determination in high-dimensional dependence learning based on
surrogate data testing, termed as inverse finite-size scaling (IFSS).
Our IFSS method is based on the discovery of a universal scaling
property of random matrices which enables inference about signal
behaviour under null hypothesis from much smaller scale surrogate data
than the dimensionality of the original data. As a motivating example,
we demonstrate the procedure for ultra high-dimensional genome data
fitted with Potts models with the order of 10^10 parameters. IFSS
reduces the computational complexity of the data testing procedure by
several orders of magnitude, making it very efficient for practical
purposes. This approach thus holds considerable potential for
generalization to other types of complex models.
|
|