Quantum-Inspired Linear Genetic Programming as a Knowledge Management System

Abstract

The superior performance of quantum computers in some problems lies in the direct use of quantum mechanics phenomena. This ability has originated the quantum-inspired evolutionary algorithms (QIEAs), which are classical algorithms (for classical computers) that exploit quantum mechanics principles to improve their performance. Several proposed QIEAs are able to outperform their traditional counterparts when applied to different kinds of problems. Aiming to exploit this new paradigm on genetic programming (GP), this paper introduces a novel QIEA model (quantum-inspired linear GP—QuaLiGP), which evolves machine code programs. QuaLiGP is inspired on multi-level quantum systems, and its operation is based on quantum individuals, which represent a superposition of all programs (solutions) of the search space. The tests use symbolic regression and binary classification as knowledge management problems to assess the QuaLiGP performance and compare it with Automatic Induction of Machine Code by Genetic Programming model, which is currently the most efficient GP model to evolve machine code. Results show that QuaLiGP outperforms the reference GP system for all these problems, by achieving better solutions from a smaller number of evaluations and by using fewer parameters and operators. This paper concludes that the quantum-inspired paradigm can be a competitive approach to evolve programs efficiently, encouraging improvements and extensions of QuaLiGP.