| Group: | Neural Networks |
| See also: | nninit nnfunc |
| Function: | nnvisu | |
| Description: | nnvisu computes the visualization for a given feed forward network by non-metric multidimensional scaling. |
| Usage: | pnew = nnvisu (d, pold) | |
| Input: | ||
| d | n x 3 matrix of distances | |
| pold | m x 2 starting position | |
| Output: | ||
| pnew | m x 2 computed new positions | |
x = read("bank2") ; read bank2 data
y =(1:200).<101
;; build logit model
wei =(1:6)~(7.*matrix(6))~normal(6)
unit = trans(#(-1,0)).*matrix(7)|trans(#(1,2))
ind = nninit(wei, unit)
wei = index(wei, ind)
randomize(0)
wei[,3] = abs(wei[,3]) ; most intersting connections are near 0
pos = normal(rows(unit),2)
pos = nnvisu(wei, pos)
pos
n/a