Learning SMaLL Predictors. (arXiv:1803.02388v1 [cs.LG])

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.

from cs updates on arXiv.org http://ift.tt/2tp3LqV
//