A Quantum-Inspired Evolutionary Algorithm Using Gaussian Distribution-Based Quantization

Abstract

Recent advances lead to the increase in the capability of evolutionary algorithms for tacking optimization problems. Particularly quantum-inspired computation leads to a new direction for enhancing the effectiveness of these algorithms. Existing studies on quantum-inspired algorithms focused primarily on evolving a single set of homogeneous solutions. This paper expands the scope of current research by applying quantum computing principles, in particular the quantum superposition principle proposing a new quantization approach. As a result, a quantum-inspired competitive coevolution algorithm (QCCEA) is proposed in this paper. Unlike its predecessors and traditional quantum-inspired algorithms, the proposed QCCEA uses a novel approach for quantization using Gaussian distribution which alters the selection procedure in order to identify the optimal fitness. This new approach which incorporates a new qubit architecture and processing has proved effective for numerical optimization problems as well as multiobjective problems. Experiments were performed on 20 benchmark numerical optimization problems as well as a combinatorial maze problem. The experiment results show that the proposed quantization technique has improved the performance and accuracy of the traditional non-quantized algorithms significantly for the both types of problems. This paper further discusses the impact of population before and after quantization as well as sensitivity of the parameters. The novel approach of quantization that improved the efficiency and performance is the key contribution of this work.