next up previous
Next: Experiments Up: The Proposed Segmentation Scheme Previous: Map Training

Threshold Determination

In the threshold determination phase each vector component j of each map unit i is assigned a threshold value tex2html_wrap_inline195 . All thresholds are determined automatically by the following procedure.

   figure70
Figure 2: The Voronoi set tex2html_wrap_inline215 of the map unit i (in a two-dimensional case).

For each map unit i the set tex2html_wrap_inline215 is defined to include only those input vectors tex2html_wrap_inline187 for which the unit i is the best-matching unit, i.e, the set tex2html_wrap_inline215 consists of those input vectors tex2html_wrap_inline187 which belong to the Voronoi set of the unit i (Figure 2),

equation78

where tex2html_wrap_inline233 is a weight vector of the map unit k. If a uniform distribution is assumed for tex2html_wrap_inline237 , the jth component of tex2html_wrap_inline241 , the mean of the tex2html_wrap_inline237 is given by

equation86

The confidence interval for tex2html_wrap_inline237 is now defined as

  equation92

where tex2html_wrap_inline247 , and d is the confidence level.

We can now define the thresholds tex2html_wrap_inline195 as

  equation105



Jukka Iivarinen
Tue Mar 5 10:03:40 EET 1996