In defense of weight-sharing for neural architecture search: an optimization perspective

Open in new window