Rotated Binary Neural Network

TL;DR

This paper introduces Rotated Binary Neural Networks (RBNN), which align weights through rotation to reduce quantization error, significantly improving performance of binary neural networks on CIFAR-10 and ImageNet.

Contribution

The paper pioneers the exploration of angular bias in BNNs and proposes a novel rotation-based method with bi-rotation matrices to enhance binarization accuracy.

Findings

RBNN achieves around 50% weight flips, maximizing information gain.

Experiments show RBNN outperforms many state-of-the-art BNN methods.

The approach effectively reduces quantization error and improves accuracy.

Abstract

Binary Neural Network (BNN) shows its predominance in reducing the complexity of deep neural networks. However, it suffers severe performance degradation. One of the major impediments is the large quantization error between the full-precision weight vector and its binary vector. Previous works focus on compensating for the norm gap while leaving the angular bias hardly touched. In this paper, for the first time, we explore the influence of angular bias on the quantization error and then introduce a Rotated Binary Neural Network (RBNN), which considers the angle alignment between the full-precision weight vector and its binarized version. At the beginning of each training epoch, we propose to rotate the full-precision weight vector to its binary vector to reduce the angular bias. To avoid the high complexity of learning a large rotation matrix, we further introduce a bi-rotation formulation that learns two smaller rotation matrices. In the training stage, we devise an adjustable rotated weight vector for binarization to escape the potential local optimum. Our rotation leads to around 50% weight flips which maximize the information gain. Finally, we propose a training-aware approximation of the sign function for the gradient backward. Experiments on CIFAR-10 and ImageNet demonstrate the superiorities of RBNN over many state-of-the-arts. Our source code, experimental settings, training logs and binary models are available at https://github.com/lmbxmu/RBNN.