Main Tab Options

    “Detection” (SD value). Threshold level is estimated as:
Thr =  SDvalue, x SDn
where SDn is an estimate of the standard deviation of the background noise. SDvalue can vary from 4 to 8. Default SDvalue is 7.

- “THRESHOLD” (New/Keep) is the control button allowing automatic (New) or manual (Keep) estimation of the threshold. In the last case you can adjust it by moving the cursor (red line) inside the “Raw signal graph” (3).

- "FCM prototype" (New/Keep) allows FCM clustering technique to make a new space segmentation and calculation of cluster’s centers (New), which is a necessary step of FCM unsupervised clustering technique. “Keep” mode permits the classification of new objects and their membership values to the different classes for the given cluster centers, that is used during classification procedure.

- “# clusters” (Auto/Manual) allows automatic determination of the number of clusters to be determined (Auto). “Manual” operation forces clustering algorithm to find exact (indicated at the left side) number of clusters. The range is from 1 to 12.

- “# features” (Auto/Manual) allows automatic determination of the number of PCs to be used during clustering/classification (Auto). “Manual” operation allows user to manually choose dimensionality by inserting the value in the field next to the button. Range is 1 to 5.

“Exponent” determines the degree of fuzziness of the resulting clustering process. As m-->inf clustering tends to be fuzzy, while m--> inf in crisp clustering. The value of 1.1 is the default.


Site Map | Printable View | © 2010 - 2019 Dr. Andriy Oliynyk

Creative Commons License
.