Abstract

Currently, deep neural networks are the state of the art on problems such as speech recognition and computer vision. We show that by using a method called model compression that shallow feed-forward nets can learn the complex functions previously learned by deep nets and achieve accuracies previously only achievable with deep models. Moreover, in some cases the shallow neural nets can learn these deep functions using the same number of parameters as the original deep models. On the TIMIT phoneme recognition and CIFAR -10 image recognition tasks, shallow nets can be trained that perform similarly to complex, well-engineered, deeper convolutional architectures. Our success in training shallow neural nets to mimic deeper models suggests that there may be better algorithms for training shallow nets than those currently available. I’ll also briefly discuss work we’re doing to compress extremely large deep models and ensembles of deep models to “modest-size” deep models that fit on servers, and to “small” deep models that run on mobile devices.