Multiclass classification using quantum convolutional neural networks with hybrid quantum-classical learning

TL;DR

This paper introduces a hybrid quantum-classical convolutional neural network approach for multiclass classification, demonstrating comparable accuracy to classical models on the MNIST dataset using quantum circuits.

Contribution

It proposes a novel quantum perceptron model and optimized quantum circuit structure for multiclass classification, advancing quantum neural network applications.

Findings

Achieved similar accuracy to classical CNNs on MNIST with quantum models.

Extended quantum classification to four classes, surpassing previous three-class results.

Demonstrated feasibility of quantum neural networks for multiclass problems in the NISQ era.

Abstract

Multiclass classification is of great interest for various applications, for example, it is a common task in computer vision, where one needs to categorize an image into three or more classes. Here we propose a quantum machine learning approach based on quantum convolutional neural networks for solving the multiclass classification problem. The corresponding learning procedure is implemented via TensorFlowQuantum as a hybrid quantum-classical (variational) model, where quantum output results are fed to the softmax activation function with the subsequent minimization of the cross entropy loss via optimizing the parameters of the quantum circuit. Our conceptional improvements here include a new model for a quantum perceptron and an optimized structure of the quantum circuit. We use the proposed approach to solve a 4-class classification problem for the case of the MNIST dataset using eight qubits for data encoding and four ancilla qubits; previous results have been obtained for 3-class classification problems. Our results show that accuracies of our solution are similar to classical convolutional neural networks with comparable numbers of trainable parameters. We expect that our finding provide a new step towards the use of quantum neural networks for solving relevant problems in the NISQ era and beyond.