An adaptive scaling technique to quantum clustering

Abstract

Data clustering is an essential tool for entering the data world. Quantum clustering (QC) is a meta-heuristic method derived from the Schrödinger equation and incorporates some concepts of quantum mechanics. QC shows a very high ability to adapt to data distribution and finding data structure. The data are clustered by the extremes of the potential function obtained from the Schrödinger equation. QC has a length parameter that plays an important task in determining the number and location of extremes of the potential function. The length parameter is the width of the Gaussian kernel that is included as a wave function in the Schrödinger equation. In this paper, we use a simple method to estimate the width of the Gaussian kernel based on an adaptive scaling technique. In adaptive scaling quantum clustering (ASQC), global hyper-parameter of QC is replaced by a local hyper-parameter. By this technique, the effects of local density are entered in QC, the necessary parameters for this model are calculated by algorithm and the clustering results are improved.