Dr. Mathieu Blondel 【15:00 - 16:00】
(Research scientist at NTT Communication Science Laboratories)
Title
Higher-Order Factorization Machines
Abstract
Factorization machines (FMs) are a supervised learning approach that can use second-order feature combinations even when the data is very high-dimensional.
Unfortunately, despite increasing interest in FMs, there exists to date no efficient training algorithm for higher-order FMs (HOFMs).
In this talk, I will present the first generic yet efficient algorithms for training arbitrary-order HOFMs.
I will also present new variants of HOFMs with shared parameters, which greatly reduce model size and prediction times while maintaining similar accuracy.
I will demonstrate the proposed approaches on four different link prediction tasks.