Quantum Convolutional Neural Network For Image Classification
Guoming Chen, Qiang Chen, Shun Long, Weiheng Zhu
Abstract
In this paper we propose a MERA inspired ternary unitary circuit feature extraction method based on Quantum Convolutional Neural Network(QCNN) in the TensorFlow quantum framework for binary image classification. The image data is properly downscaled with the ternary unitary circuits before fed into the QCNNs quantum circuit for state preparation, quantum convolution and quantum pooling. High or appropriate entangled state corresponds to the high separated weight function. When entanglement is reduced, the classification result can then be directly obtained from the qubit. We trained quantum classifiers with QCNN and with two hybrid quantum-classical QCNN models and their performance are compared against that of a classic CNN. The results on the breast cancer image dataset show that the proposed QCNN with the proposed feature extraction outperformed the classic CNN in terms of recognition rate 90%-93% over 83%-85% on performance.