2019 (modified: 11 Nov 2022)ICML 2019Readers: Everyone
Abstract:We introduce the concrete autoencoder, an end-to-end differentiable method for global feature selection, which efficiently identifies a subset of the most informative features and simultaneously le...