Binarized Neural Networks

Authors: Itay Hubara, Matthieu Courbariaux, Daniel Soudry, Ran El-Yaniv, Yoshua Bengio

NeurIPS 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental To validate the effectiveness of BNNs, we conducted two sets of experiments on the Torch7 and Theano frameworks.
Researcher Affiliation Academia (1) Technion, Israel Institute of Technology. (2) Université de Montréal. (3) Columbia University. (4) CIFAR Senior Fellow.
Pseudocode Yes Algorithm 1: Training a BNN.
Open Source Code Yes The code for training and running our BNNs is available on-line (both Theano1 and Torch framework2).
Open Datasets Yes On both, BNNs achieved nearly state-of-the-art results over the MNIST, CIFAR-10 and SVHN datasets.
Dataset Splits Yes Figure 1: Training curves for different methods on CIFAR-10 dataset. The dotted lines represent the training costs (square hinge losses) and the continuous lines the corresponding validation error rates.
Hardware Specification Yes The first three columns represent the time it takes to perform a 8192 8192 8192 (binary) matrix multiplication on a GTX750 Nvidia GPU, depending on which kernel is used.
Software Dependencies Yes Torch7: A matlab-like environment for machine learning.
Experiment Setup No Implementation details are reported in Appendix A and code for both frameworks is available online. However, these specific details (e.g., hyperparameters like learning rate, batch size) are not provided in the main body of the paper.