Sketching and Neural Networks

Amit Daniely, Nevena Lazic, Yoram Singer +1

High-dimensional sparse data present computational and statistical challenges for supervised learning. We propose compact linear sketches for reducing the dimensionality of the inp…