BinaryNet: Training Deep Neural Networks with Weights and Activations Constrained to +1 or -1
Year: 2016
Type: preprint
Abstract: We introduce BinaryNet, a method which trains DNNs with binary weights andactivations when computing parameters' gradient. We show that it is possible to
train a Multi Layer Perceptron (MLP) on MNIST and ConvNets on CIFAR-10 and SVHN
with BinaryNet and achieve nearly state-of-the-art results. At run-time,
BinaryNet drastically reduces memory usage and replaces most multiplications by
1-bit exclusive-not-or (XNOR) operations, which might have a big impact on both
general-purpose and dedicated Deep Learning hardware. We wrote a binary matrix
multiplication GPU kernel with which it is possible to run our MNIST MLP 7
times faster than with an unoptimized GPU kernel, without suffering any loss in
classification accuracy. The code for BinaryNet is available.
Source: arXiv (Cornell University)
Authors: Matthieu Courbariaux, Yoshua Bengio
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2260663238