Home > Research > Publications & Outputs > Quantum Face Recognition with Multi-Gate Quantu...

Electronic data

  • MG-QNN_TAI_r3_v6.0

    Accepted author manuscript, 944 KB, PDF document

    Available under license: CC BY: Creative Commons Attribution 4.0 International License

Links

Text available via DOI:

View graph of relations

Quantum Face Recognition with Multi-Gate Quantum Convolutional Neural Network

Research output: Contribution to Journal/MagazineJournal articlepeer-review

Published
Close
<mark>Journal publication date</mark>31/12/2024
<mark>Journal</mark>IEEE Transactions on Artificial Intelligence
Issue number12
Volume5
Number of pages12
Pages (from-to)6330-6341
Publication StatusPublished
Early online date28/06/24
<mark>Original language</mark>English

Abstract

In the last decade, quantum computing has showcased its unique mechanism across diverse fields, highlighting significant potential for data-driven applications requiring substantial computational resources. Within this landscape, quantum machine learning emerges as a promising frontier, poised to harness the unique advantages of quantum computing for machine learning tasks. Nonetheless, the current generation of quantum hardware, typified by noisy intermediatescale quantum (NISQ) devices, grapples with severe resource constraints, particularly in terms of qubit availability. While quantum computing offers tantalizing capabilities such as superposition and entanglement, which can be strategically leveraged to optimize the performance of quantum neural networks, the challenge remains in mitigating the resource limitations while upholding high recognition accuracy. To address this imperative, we introduce a pioneering face recognition method christened the Multi-Gate Quantum Convolutional Neural Network (MG-QCNN). This innovation is engineered to surmount the resource bottleneck endemic to NISQ devices while preserving exceptional recognition accuracy. Our empirical investigations conducted on benchmark datasets, including the Yale face dataset and the ORL face database, illuminate the remarkable potential of this approach. Specifically, our proposed variational quantum circuit architecture consistently achieves an impressive average accuracy of 96%, which is better than the 95% of the classic CNN. Our model underscores the efficacy of quantum convolution operations in the extraction of feature maps, exhibiting a transformative stride toward unlocking the full potential of quantum-enhanced face recognition, and compared with other quantum models, our method has more advantages in accuracy and efficiency.