Bitwise Neural Networks

TL;DR

This paper introduces Bitwise Neural Networks (BNNs), which use binary weights, biases, and signals to enable efficient, low-resource neural network deployment suitable for hardware with limited capacity.

Contribution

The paper proposes a novel training process for BNNs, including weight compression and noisy backpropagation, achieving near real-valued network performance with binary operations.

Findings

BNNs perform competitively on MNIST with binary features.

BNNs significantly reduce computational complexity and power consumption.

BNNs are suitable for resource-constrained environments.

Abstract

Based on the assumption that there exists a neural network that efficiently represents a set of Boolean functions between all binary inputs and outputs, we propose a process for developing and deploying neural networks whose weight parameters, bias terms, input, and intermediate hidden layer output signals, are all binary-valued, and require only basic bit logic for the feedforward pass. The proposed Bitwise Neural Network (BNN) is especially suitable for resource-constrained environments, since it replaces either floating or fixed-point arithmetic with significantly more efficient bitwise operations. Hence, the BNN requires for less spatial complexity, less memory bandwidth, and less power consumption in hardware. In order to design such networks, we propose to add a few training schemes, such as weight compression and noisy backpropagation, which result in a bitwise network that performs almost as well as its corresponding real-valued network. We test the proposed network on the MNIST dataset, represented using binary features, and show that BNNs result in competitive performance while offering dramatic computational savings.