Your data is woefully inadequate. Let
[ I N ] = size(input)
[ O N ] = size(target)
Ntrn = N-2*round(0.15*N) % default training set size
Ntrneq = N*O % default number of training equations
For an adequate definition of the input and output spaces, it is desirable that
Ntrn >> 1+max(I,O)
For a net with I-H-O node topology, the number of unknown weights is
Nw = (I+1)*H+(H+1)*O
For an adequate estimate of weights, it is desirable that
Ntrneq >> Nw
You need a drastic reduction in the input dimension I and/or a corresponding increase in training examples.
When Ntrn is not large and/or the number of classification categories is large,pay more attention to the function confusion than the plot function plotconfusion.