出版社：Vilnius University, University of Latvia, Latvia University of Agriculture, Institute of Mathematics and Informatics of University of Latvia
摘要：In the paper, the influence of learning parameters on self-organizing map (SOM) is analyzed, when both numerical data and text documents are investigated. Three neighboring functions (bubble, Gaussian, and heuristic) and four learning rates (linear, inverse-of-time, power series, and heuristic) have been investigated. The learning rates are changed according to epochs or iterations. The quality of self-organizing map is measured not only by quantization error, but also by two other measures, which are suitable when the classified data are analyzed.
关键词：self-organizing map; neighboring function; learning rate; text document matrix; SOM quality estimators.
