Hybrid Classical Quantum Neural Network based classification of Photonic Band Gap crystal structure defects
P. Paayas, S. Sridevi, Thirunavukkarasu Kanimozhi, M. Valliammai, Jayavelu Mohanraj, Vinodh Kumar N, A. Bakiya, R Prasanna Kumar, M Rithani · 2023
This paper emphasizes an effort to effectively identify and categorize structural defects in Photonic Band Gap (PBG) Crystals, which are vital in material characterization and structure analysis. These PBG materials have unique optical characteristics that are quite advantageous in the design of photonic devices and solar cells however, their imperfections may exert an adverse impact on these characteristics. The challenging task of recognizing and classifying these PBG crystal defects with high accuracy and efficiency can be handled by using machine learning techniques. We propose the use of a Hybrid Classical Quantum Neural Network (HCQNN) model that leverages the traits of quantum computing to efficiently classify the types of structural defects in the PBG crystals. The quantum part of the HCQNN model is contrived with a quantum node encompassing three qubits, Angle Embedding layer and Basic entangling layer with 500 numbers of shots. The quantum node interface is integrated with generalized classical dense layers, and the entire HCQNN model is trained with a classical optimizer. The experimental results demonstrate that the proposed HCQNN architecture outperforms existing machine learning and Artificial Neural Network models in terms of classification accuracy, precision, and other critical model parameters.