Parameter Estimation Method for Weighted Euclidean Distance Model

Abstract

This paper describes a new parameter estimation method for weighted euclidean distance model, in which individual configurations obtained from multidimensional scaling are directly fitted to some common space. The new method is based on an alternating least squares estimation method, and is easily extended for more general weighted euclidean model. Validity of the method is illustrated by using and analyzing a set of artificial data and color perception data.