Paper ID: 2412.02059

Lean classical-quantum hybrid neural network model for image classification

Ao Liu, Cuihong Wen, Jieci Wang

The integration of algorithms from quantum information with neural networks has enabled unprecedented advancements in various domains. Nonetheless, the application of quantum machine learning algorithms for image classiffcation predominantly relies on traditional architectures such as variational quantum circuits. The performance of these models is closely tied to the scale of their parameters, with the substantial demand for parameters potentially leading to limitations in computational resources and a signiffcant increase in computation time. In this paper, we introduce a Lean Classical-Quantum Hybrid Neural Network (LCQHNN), which achieves efffcient classiffcation performance with only four layers of variational circuits, thereby substantially reducing computational costs. We apply the LCQHNN to image classiffcation tasks on public datasets and achieve a classiffcation accuracy of 99.02% on the dataset, marking a 5.07% improvement over traditional deep learning methods. Under the same parameter conditions, this method shows a 75% and 70.59% improvement in training convergence speed on two datasets. Furthermore, through visualization studies, it is found that the model effectively captures key data features during training and establishes a clear association between these features and their corresponding categories. This study conffrms that the employment of quantum algorithms enhances the model's ability to handle complex classiffcation problems.

Submitted: Dec 3, 2024