Abstract | ||
---|---|---|
Self-organizing maps (SOM) had been used for input data quantization and visual display of data, an important property that does not exist in most of clustering algorithms. Effective data clustering using SOM involves two or three steps procedure. After proper network training, units can be clustered generating regions of neurons which are related to data clusters. The basic assumption relies on the data density approximation by the neurons through unsupervised learning. This paper presents a gradient-based SOM visualization method and compares it with U-matrix. It also discusses steps toward clustering using SOM and morphological operators. Results using benchmark datasets show that the new method is more robust to choice of parameters in the filtering phase than the conventional method. The paper also proposes an enhancing method to map visualization taking advantage of the neurons activity, which improve cluster detection especially in small maps. |
Year | DOI | Venue |
---|---|---|
2010 | 10.1007/978-3-642-15381-5_41 | IDEAL |
Keywords | Field | DocType |
input data quantization,gradient-based som visualization method,conventional method,data cluster,steps procedure,neurons activity,effective data,data density approximation,new method,clustering algorithm,visualizing som result,neuronal activity,unsupervised learning,visualization,data clustering,self organizing maps | Data mining,Fuzzy clustering,CURE data clustering algorithm,Computer science,Self-organizing map,Consensus clustering,Artificial intelligence,Conceptual clustering,Cluster analysis,Canopy clustering algorithm,Pattern recognition,Correlation clustering,Machine learning | Conference |
Volume | ISSN | ISBN |
6283 | 0302-9743 | 3-642-15380-1 |
Citations | PageRank | References |
5 | 0.48 | 12 |
Authors | ||
1 |
Name | Order | Citations | PageRank |
---|---|---|---|
José Alfredo F. Costa | 1 | 52 | 10.11 |