Home » Multiple Hypothesis Testing with Persistent Homology

Multiple Hypothesis Testing with Persistent Homology

SEP 27, 2019 | 11:45 AM TO 12:30 PM

Details

WHERE: The Graduate Center
365 Fifth Avenue
ROOM: 3209
WHEN: Sep 27, 2019: 11:45 AM – 12:30 PM
CONTACT INFO: Applied Topology Seminar
ADMISSION: Free
SPONSOR:
Data Science and Applied Topology Seminar

 

Description

Mikael Vejdemo-Johansson will speak about recent research into Multiple hypothesis testing with persistent homology.

Abstract: Based on joint work with Sayan Mukherjee, we will talk about the need for multiple hypothesis corrections – both for Family-Wise Error Rates (decrease the risk of even a single false positive) and for False Discovery Rates (fix the rate of false positives at a chosen level) and how to correct both types in hypothesis testing using persistent homology.