Benchmarking Quantum Convolutional Neural Networks for Classification and Data Compression Tasks

TL;DR

This paper evaluates the performance of Quantum Convolutional Neural Networks (QCNNs) in quantum state classification and data compression, demonstrating their efficiency and comparable accuracy to hardware-efficient ansatz models across different system sizes.

Contribution

It provides a comparative analysis of QCNNs and HEAs for quantum state classification and compression, highlighting QCNNs' faster training with fewer parameters.

Findings

QCNNs with RY gates train faster than HEAs.

QCNNs match HEA performance in classification tasks.

QCNNs effectively compress quantum states.

Abstract

Quantum Convolutional Neural Networks (QCNNs) have emerged as promising models for quantum machine learning tasks, including classification and data compression. This paper investigates the performance of QCNNs in comparison to the hardware-efficient ansatz (HEA) for classifying the phases of quantum ground states of the transverse field Ising model and the XXZ model. Various system sizes, including 4, 8, and 16 qubits, through simulation were examined. Additionally, QCNN and HEA-based autoencoders were implemented to assess their capabilities in compressing quantum states. The results show that QCNN with RY gates can be trained faster due to fewer trainable parameters while matching the performance of HEAs.