Quantum -Inspired Evolutionary Algorithm for Hyperparameter Tuning in Automatic Number Plate Recognition Systems

Hyperparameter tuning plays a pivotal role in optimizing machine learning models, especially in complex tasks such as Automatic Number Plate Recognition (ANPR). This paper presents a Quantum -Inspired Evolutionary Algorithm (QEA) for tuning hyperparameters of a deep learning -based ANPR system tailored to Persian license plates. Unlike traditional approaches such as grid or random search, QEA simulates quantum phenomena like superposition and entanglement on classical hardware to enhance exploration in high -dimensional search spaces. Our experiments, conducted on a publicly available dataset using TensorFlow and Qiskit frameworks, reveal that QEA achieves superior validation accuracy (۸۷%) and faster convergence compared to classical evolutionary algorithms (۸۲%) and Bayesian optimization (۸۴%). We also detail implementation procedures, preprocessing techniques, and discuss the algorithm's scalability and robustness to real -world noise such as varied fonts and lighting. The proposed method shows practical viability in real -world ANPR applications like traffic monitoring and toll collection. Code is available on GitHub.

کلیدواژه ها:

Quantum -Inspired Optimization ، Hyperparameter Tuning ، Automatic Number Plate Recognition