A trial of using cluster analysis for classifying ship noise and electroencephalogram

In multi-dimensional characteristic space a sample can be represented by its characteristic values. After nonlinear mapping proposed by Sammon, J. W. in 1969, in lower dimensional space of characteristics, samples with different characteristic values will be easily classified. On purpose to prove that closter analysis is suitable for quite different kinds of samples, in this paper some ship noises and some E E G samples are classified and shown. And it is worthy to point out that the adaptive step size expression of adaptive iteration deduced here could also be effective if it was applied to speed up convergence of adaptive algorithm used for signal processing.