Stereoscopic scalable quantum convolutional neural networks.

Area of Science:

• Quantum Computing
• Machine Learning
• Artificial Intelligence

Background:

• The noisy intermediate-scale quantum (NISQ) era necessitates advanced quantum algorithms.
• Quantum neural networks (QNNs) and quantum convolutional neural networks (QCNNs) show promise for complex problems.
• Scaling QCNNs is hindered by barren plateaus, especially for high-dimensional data classification.

Purpose of the Study:

• To propose a novel stereoscopic 3D scalable QCNN (sQCNN-3D) for point cloud data processing.
• To enhance feature diversity in QCNNs using limited qubits via reverse fidelity training (RF-Train).
• To address the challenges of scaling QCNNs and improving classification performance.

Main Methods:

• Development of a stereoscopic 3D scalable QCNN (sQCNN-3D).
• Integration of reverse fidelity training (RF-Train) with sQCNN-3D.
• Performance evaluation using data-intensive methods on point cloud datasets.

Main Results:

• The proposed sQCNN-3D effectively processes high-dimensional point cloud data.
• RF-Train enhances feature diversification, overcoming limitations of a small number of qubits.
• The combined approach achieves desired performance in classification applications.

Conclusions:

• The sQCNN-3D with RF-Train offers a viable solution for high-dimensional data classification in quantum computing.
• This method addresses the barren plateau challenge in QCNN scaling.
• The research contributes to advancing quantum machine learning for complex data processing.