he04ijcnn
Ji He, Man Lan, Chew-Lim Tan, Sam-Yuan Sung and Hwee-Boon Low. “Initialization of Cluster Refinement Algorithms: A Review and Comparative Study”. In the Proceedings of International Joint Conference on Neural Networks (IJCNN). July 2004. p297-302. Also here: Supplementary resources of this paper.
Supplementary Resource for HE04IJCNN
A) The Gaussian Random Generator (C++ Source Code)
B) The Sythetic Data Sets Used in Our Experiments
B.1) The Designated Cluster Centroids
01: (0.070000, 0.082000)
02: (0.288000, 0.120000)
03: (0.687000, 0.066000)
04: (0.522000, 0.198000)
05: (0.782000, 0.190000)
06: (0.258000, 0.310000)
07: (0.138000, 0.462000)
08: (0.438000, 0.460000)
09: (0.692000, 0.414000)
10: (0.458000, 0.588000)
11: (0.622000, 0.582000)
12: (0.302000, 0.658000)
13: (0.184000, 0.870000)
14: (0.608000, 0.859000)
15: (0.856000, 0.752000)
By default the above Gassian random generator program writes the designated N cluster centroids as the first N instances in the output file (i.e. line 3 - line N + 2).
B.2) The Three Data Sets Reported in the Manuscript
S01 S13 S25 (each one about 1M zipped file)
B.3) The EPS Images of the Sythetic Data Sets
Separated EPS file, each corresponding to one data set (625K zipped file)
The overview of the full collection (219K EPS file)
1% data points (i.e. 1,500 out of 150,000) uniformly randomly selected from each data set are used to plot the corresponding EPS file.
C) The Complete Experimental Results