Training Binary Multilayer Neural Networks for Image Classification using Expectation Backpropagation

TL;DR

This paper evaluates the effectiveness of Binary Multilayer Neural Networks trained with Expectation Backpropagation on image classification, demonstrating competitive accuracy on MNIST with binary weights.

Contribution

It extends BMNNs to multiclass image tasks and explores techniques like spatial filters and dropout, showing competitive results with standard methods.

Findings

EBP achieves 2.12% test error with binary weights

EBP achieves 1.66% test error with real weights

BMNNs perform comparably to standard neural networks on MNIST

Abstract

Compared to Multilayer Neural Networks with real weights, Binary Multilayer Neural Networks (BMNNs) can be implemented more efficiently on dedicated hardware. BMNNs have been demonstrated to be effective on binary classification tasks with Expectation BackPropagation (EBP) algorithm on high dimensional text datasets. In this paper, we investigate the capability of BMNNs using the EBP algorithm on multiclass image classification tasks. The performances of binary neural networks with multiple hidden layers and different numbers of hidden units are examined on MNIST. We also explore the effectiveness of image spatial filters and the dropout technique in BMNNs. Experimental results on MNIST dataset show that EBP can obtain 2.12% test error with binary weights and 1.66% test error with real weights, which is comparable to the results of standard BackPropagation algorithm on fully connected MNNs.