Quaternion Convolutional Neural Networks: Current Advances and Future Directions

TL;DR

This paper reviews the development of Quaternion-Valued CNNs, highlighting their richer representations and parameter efficiency, and discusses future research directions in this emerging field.

Contribution

It provides a comprehensive organization of current QCNN research and proposes future directions for advancing quaternion-based neural networks.

Findings

QCNNs can achieve similar performance with fewer parameters.

Hyper-complex numbers offer richer representational capacities.

Hamilton products capture intrinsic interchannel relationships.

Abstract

Since their first applications, Convolutional Neural Networks (CNNs) have solved problems that have advanced the state-of-the-art in several domains. CNNs represent information using real numbers. Despite encouraging results, theoretical analysis shows that representations such as hyper-complex numbers can achieve richer representational capacities than real numbers, and that Hamilton products can capture intrinsic interchannel relationships. Moreover, in the last few years, experimental research has shown that Quaternion-Valued CNNs (QCNNs) can achieve similar performance with fewer parameters than their real-valued counterparts. This paper condenses research in the development of QCNNs from its very beginnings. We propose a conceptual organization of current trends and analyze the main building blocks used in the design of QCNN models. Based on this conceptual organization, we propose future directions of research.