no code implementations • NeurIPS 2018 • Matthew Olson, Abraham Wyner, Richard Berk
In this paper, we use a linear program to empirically decompose fitted neural networks into ensembles of low-bias sub-networks.
General Classification Small Data Image Classification