Winning the Lottery with Continuous Sparsification

    Pedro Savarese           Hugo Silva            Michael Maire
    TTI-Chicago           University of Alberta       University of Chicago
  savarese@ttic.edu         hugoluis@ualberta.ca        mmaire@uchicago.edu

                         Abstract
       The search for efficient, sparse deep neural network models is most prominently
       performed by pruning: training a dense, overparameterized network and removing
       parameters, us ...