AURUM Mühendislik Sistemleri ve Mimarlık Dergisi

Yazarlar: Amna Mohamed M. ABURAS, Wamidh MAZHER, Osman Nuri UÇAN, Oğuz BAYAT

Konular:-

Anahtar Kelimeler:Clustering,K-means,Spectral method,Laplace,Eigenvector,GUI

Özet: Abstract Spectral clustering is developed for both normalized and unnormalized methods. However, selecting between the two methods is not established in the GUI (Graphical User Interface) yet . In this paper , we implement different clustering algorithms using GUI-MATLAB, then, the clustering by these three methods, is compared for similar pairs of datasets. Our model is employing such three different clustering methods which are spectral, hierarchical and density based methods, then employing different geometrical, multi-range, and multi-level similar datasets pairs of graph for clustering. As result, the above three clustering algorithms are experimented for different environments which are (geometrical, multi-range and multi-level). The simulation result shows the clustering of these pairs of geometrical datasets which are: Concentric circles, Semi-circles, and Aggregation. Accordingly, the spectral algorithm has superior clustering in case of big datasets more than 2000 pairs points and range more than 500 levels among datasets

---