Quantum-inspired algorithm for direct multi-class classification

GIUNTINI, ROBERTO; HOLIK, FEDERICO; PARK, DANIEL K.; FREYTES, HECTOR; BLANK, CARSTEN; SERGIOLI, GIUSEPPE

APPLIED SOFT COMPUTING

ELSEVIER SCIENCE BV

Año: 2023 vol. 134

1568-4946

Over the last few decades, quantum machine learning has emerged as a groundbreaking discipline.Harnessing the peculiarities of quantum computation for machine learning tasks offers promisingadvantages. Quantum-inspired machine learning has revealed how relevant benefits for machinelearning problems can be obtained using the quantum information theory even without employingquantum computers. In the recent past, experiments have demonstrated how to design an algorithmfor binary classification inspired by the method of quantum state discrimination, which exhibits highperformance with respect to several standard classifiers. However, a generalization of this quantuminspired binary classifier to a multi-class scenario remains nontrivial. Typically, a simple solutionin machine learning decomposes multi-class classification into a combinatorial number of binaryclassifications, with a concomitant increase in computational resources. In this study, we introducea quantum-inspired classifier that avoids this problem. Inspired by quantum state discrimination, ourclassifier performs multi-class classification directly without using binary classifiers. We first comparedthe performance of the quantum-inspired multi-class classifier with eleven standard classifiers. Thecomparison revealed an excellent performance of the quantum-inspired classifier. Comparing theseresults with those obtained using the decomposition in binary classifiers shows that our methodimproves the accuracy and reduces the time complexity. Therefore, the quantum-inspired machinelearning algorithm proposed in this work is an effective and efficient framework for multi-classclassification. Finally, although these advantages can be attained without employing any quantumcomponent in the hardware, we discuss how it is possible to implement the model in quantumhardware.