the success probability for the reclassification criterion, as
described (previous results). Approximating this distri-
bution to the values of U; such that lUl p 1; it is possible
to estimate the probability of classification error without
using a set of tests.
In the results obtained, in only one sample no class was
indicated (all the outputs were negative—Table 1), and
the numerical difference between the outputs of the class
indicated (positive) and the other classes (negative), resulted
in a distribution of PðUÞ far from zero for all the classes.
Consequently, the probability of success for all the classes is
approximately 100.0%. An example of distribution obtained
for the UC class is shown in Fig. 4, the others not being
presented due to their similarity. A more in-depth
calculation is found in Ref. [14].
the success probability for the reclassification criterion, as
described (previous results). Approximating this distri-
bution to the values of U; such that lUl p 1; it is possible
to estimate the probability of classification error without
using a set of tests.
In the results obtained, in only one sample no class was
indicated (all the outputs were negative—Table 1), and
the numerical difference between the outputs of the class
indicated (positive) and the other classes (negative), resulted
in a distribution of PðUÞ far from zero for all the classes.
Consequently, the probability of success for all the classes is
approximately 100.0%. An example of distribution obtained
for the UC class is shown in Fig. 4, the others not being
presented due to their similarity. A more in-depth
calculation is found in Ref. [14].
การแปล กรุณารอสักครู่..