|
Title:
|
ROUGH SET CLUSTERING APPROACH USING NON-DOMINATED SORTING GENETIC ALGORITHM |
|
Author(s):
|
Tansel Özyer , Reda Alhajj , Ken Barker |
|
ISBN:
|
972-98947-3-6 |
|
Editors:
|
Nuno Guimarães and Pedro Isaías |
|
Year:
|
2004 |
|
Edition:
|
Single |
|
Keywords:
|
Rough Sets, Clustering, NSGA-II, Cluster Validity Index. |
|
Type:
|
Full Paper |
|
First Page:
|
1139 |
|
Last Page:
|
1146 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
The purpose of the work described in this paper is estimating the number of clusters for web data by using a multi-objective evolutionary algorithm that employs a fast non-dominated sorting genetic algorithm - NSGA-II. Traditional methods scalarize the multi-objective vector into a single objective vector giving different weight values to each of the objective vectors. The user may aim to find the pareto-optimal front set of solutions. Multi-objective optimization has an overhead such as predicting the weight of each objective vector, besides unsupervised classification methods need to know the number of clusters required. It is difficult to predict the number of clusters even for the domain experts. By applying a heuristic, among those solutions, feasible solutions are chosen for each cluster and minimum cluster satisfying our heuristic is chosen as the number of clusters in the result. The answer may not be one and our system may find more than one solution for a specific cluster number. The domain expert can analyze the result and take a decision. This will also help the expert figure out the possible results available. In order to demonstrate the applicability of our method, we used the website one of the course at the University of Calgary in the experiments. |
|
|
|
|
|
|
