Abstract
This paper introduces two innovative algorithms: the State Vector Classification Algorithm (SVCA), inspired by quantum principles, and its hybrid counterpart, the Quantum State Vector Classification Algorithm (QSVCA). Both algorithms operate by mapping input data into a classification space and projecting it onto the basis vector with the largest component. This projection operation serves as the central for classifying elements within a dataset and introducing non-linearity without relying on an activation function. Specifically, the SVCA’s classification process bears resemblance to measurements in quantum computing, facilitating a seamless transition of many SVCA processes into a quantum framework. Furthermore, extensive evaluation results show that the SVCA performs comparably with other classical classifiers and notably enhances accuracy on the tested elliptical dataset. Additionally, the QSVCA serves as a quantum counterpart to the classical SVCA, consistently demonstrating strong performance on simple relationships within datasets and even outperforming the Linear Support Vector Machine SVM and the Random Forest Classifier on a synthetic dataset. Both algorithms maintain consistent performance across multiple datasets, highlighting their reliability.
