2-D normal cloud model based on fitting
Jin Gou · Jisuanji gongcheng yu sheji · 2009
A new algorithm of forward and backward cloud based on fitting is proposed. Firstly, whole concept space is divided into several sub-spaces, and backward or foreword cloud generator algorithms on each sub-space is used to fit the representation of concept. During the procedure of fitting, the number of sub-space is the major factor to fitting results. After a series of experiments, an advisable number of sub-spaces are given. After a comparison is made between these two kinds of algorithms, the conclusion that cloud model based on fitting is better for describing uncertainty than the older one could be drawn.