Learning SMaLL Predictors

03/06/2018

∙

by   Vikas K. Garg, et al.

∙

0

∙

share

We present a new machine learning technique for training small resource-constrained predictors. Our algorithm, the Sparse Multiprototype Linear Learner (SMaLL), is inspired by the classic machine learning problem of learning k-DNF Boolean formulae. We present a formal derivation of our algorithm and demonstrate the benefits of our approach with a detailed empirical study.