Polynomial Networks and Factorization Machines: New Insights and Efficient Training Algorithms

TL;DR

This paper unifies polynomial networks and factorization machines under a new perspective, introduces efficient training algorithms via low-rank tensor estimation, and demonstrates their effectiveness on regression and recommender system tasks.

Contribution

It provides a unified framework for polynomial networks and factorization machines and proposes novel efficient training algorithms based on low-rank tensor estimation.

Findings

Effective training algorithms for both models

Improved performance on regression tasks

Enhanced recommender system results

Abstract

Polynomial networks and factorization machines are two recently-proposed models that can efficiently use feature interactions in classification and regression tasks. In this paper, we revisit both models from a unified perspective. Based on this new view, we study the properties of both models and propose new efficient training algorithms. Key to our approach is to cast parameter learning as a low-rank symmetric tensor estimation problem, which we solve by multi-convex optimization. We demonstrate our approach on regression and recommender system tasks.