|
Activity Number:
|
529
- SPEED: Machine Learning
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 2, 2017 : 10:30 AM to 11:15 AM
|
|
Sponsor:
|
Section on Statistical Learning and Data Science
|
|
Abstract #325305
|
|
|
Title:
|
Group Fused Multinomial Regression
|
|
Author(s):
|
Brad Price* and Adam Rothman and Charles Geyer
|
|
Companies:
|
West Virginia University and University of Minnesota and University of Minnesota
|
|
Keywords:
|
Multinomial Regression ;
Fusion Penalties ;
Statistical Learning ;
ADMM ;
Model Selection
|
|
Abstract:
|
We propose a penalized likelihood method to reduce the number of response cat- egories in multinomial logistic regression. An l2 fusion penalty is used to introduce shrinkage and exploit vectorwise similarity of the regression coefficients. An ADMM optimization algorithm is used to compute the estimates, and its convergence is established. Prediction and model selection are addressed.
|
Authors who are presenting talks have a * after their name.
