期刊名称：International Journal of Computer and Information Technology
印刷版ISSN：2279-0764
出版年度：2016
卷号：5
期号：4
页码：359-365
出版社：International Journal of Computer and Information Technology
摘要：This paper presents an arithmetic affine robust
method to improve the performance of the self-organizing feature
map which further preserves the similarities between data inputs
and the weights matrix. The method presented herein targets the
pre-processing and validation steps in the iterative process by
filtering sensory uncertainties ensuing in data inaccuracy and
large standard deviation affecting cluster affinity. The method
introduces tolerances on incoming inputs to mitigate insignificant
clustering creating computational burden and biasing the end
result embedded in the topological map. The new technique
utilizes mathematical means to modify both the competitive and
adaptive stages of the conventional self-organizing map. To test
the new algorithm, a simulation study was conducted to cluster
Fisher's Iris dataset to improve the performance and robustness
of the resulting map.
