In the presented work, a hybrid neural network is proposed, which combines quantum and classical computing, and is intended for use in image classification. The hybrid neural network is implemented on the basis of a classical convolutional neural network using a quantum circuit. Also, within the framework of this study, a comparison was made of various configurations of a hybrid neural network in which a different number of qubits were used. The hybrid neural network configurations were trained and tested on the CIFAR10 and CIFAR100 datasets. Comparison of the performance of a hybrid neural network for multiclass classification was carried out for a different number of classes (from 2 to 10) with the corresponding number of qubits (from 2 to 4). The results obtained during the experiments confirmed the possibility of using a hybrid neural network to solve the problem of multiclass classification.
Keywords: machine learning, deep learning, quantum machine learning, quantum computing, hybrid neural network, image classification, convolutional neural network, quantum circuit