Results (
Thai) 1:
[Copy]Copied!
tructure, using two optional kernels: (a) linear and (b)
RBF. Square and circle data points represent the two
classes to be classified (black objects denote the respective
support vectors of each class).
In the testing phase, by inputting an unknown pattern to
trained SVM model, we obtain an output associating this
pattern to one of the two trained classes. High outputs
indicate that the pattern likely belongs to the positive class.
Similarly, low output scores associate the unknown pattern
to the negative class. Binary (yes/no) decisions are
obtained simply by thresholding the output scores. The
strength of the score can be associated with the probability
which the classified pattern belongs to either class, displaying
this as a likelihood ratio.
Being translated, please wait..