Quantum convolutional neural network for classical data classification

TL;DR

This paper benchmarks quantum convolutional neural networks (QCNNs) inspired by classical CNNs for classifying datasets like MNIST, demonstrating high accuracy with shallow, NISQ-compatible quantum circuits and outperforming classical CNNs under similar conditions.

Contribution

It introduces a fully parameterized, two-qubit interaction QCNN model optimized for classical data classification and compares its performance with classical CNNs on standard datasets.

Findings

QCNN achieved high classification accuracy with few parameters.

QCNN outperformed classical CNNs under similar training conditions.

The model is suitable for implementation on NISQ devices.

Abstract

With the rapid advance of quantum machine learning, several proposals for the quantum-analogue of convolutional neural network (CNN) have emerged. In this work, we benchmark fully parameterized quantum convolutional neural networks (QCNNs) for classical data classification. In particular, we propose a quantum neural network model inspired by CNN that only uses two-qubit interactions throughout the entire algorithm. We investigate the performance of various QCNN models differentiated by structures of parameterized quantum circuits, quantum data encoding methods, classical data pre-processing methods, cost functions and optimizers on MNIST and Fashion MNIST datasets. In most instances, QCNN achieved excellent classification accuracy despite having a small number of free parameters. The QCNN models performed noticeably better than CNN models under the similar training conditions. Since the QCNN algorithm presented in this work utilizes fully parameterized and shallow-depth quantum circuits, it is suitable for Noisy Intermediate-Scale Quantum (NISQ) devices.