Activity Number:
|
332
- SPEED: Section on Bayesian Statistical Science
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, August 1, 2017 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Bayesian Statistical Science
|
Abstract #323325
|
View Presentation
|
Title:
|
Multi-Way Interacting Regression via Factorization Machines
|
Author(s):
|
Mikhail Yurochkin* and Long Nguyen and Nikolaos Vasiloglou
|
Companies:
|
and University of Michigan and Infor
|
Keywords:
|
Factorization Machines ;
Selection of Interactions ;
Regression ;
Prediction ;
Indian Buffet Process ;
Hypergraph
|
Abstract:
|
We propose a Bayesian nonparametric regression method that accounts for multi-way interactions of arbitrary orders among the predictor variables. Our model makes use of a factorization mechanism for representing the regression coefficients of interactions among the predictors, while the interaction selection is guided by a nonparametric prior distribution on random hypergraphs, a construction using the Indian Buffet Process as a building block. We present a posterior inference algorithm based on Gibbs sampling, and establish posterior consistency of our regression model. Our method is evaluated with extensive experiments on simulated data and demonstrates superior performance in several applications in genetics and retail demand forecasting.
|
Authors who are presenting talks have a * after their name.