T Cell–Inspired Algorithms - Immunological Computation: Theory and Applications - page 99

Hyper-ellipse-shaped detector if the closest point of the sample hypersphere
to the center of the hyperellipse is inside it as shown in Figure 4.17 (Shapiro
et al., 2005).

A number of runs of the evolutionary algorithm are required for a given self-set
to generate a population of feasible detectors to cover the nonself space. h en, the
best detector is selected to be added to the detector set, D . Figure 4.18 presents a
pseudocode for the “evolutionary detector generation.”
Each time a new detector is added to the detector list, the list is sorted in descend-
ing order based on coverage. h e coverage is computed using a Monte Carlo esti-
mation as follows: each detector is evaluated against a sequence of random points
uniformly distributed in [0, 1] n to measure the percentage covered; the matched
points are subsequently removed from the list. If any of the current detectors has
an eff ective coverage of zero, it is assumed that the contribution of this detector is
negligible and it is thus eliminated from the detector list.

Ellipse

P 2 − P 1
P 2 − P 1

P 1 P 3 = P 2 − r

P 3

r

Self hypersphere

P 2

Figure 4.17 Determining the overlap of a hyperellipse detector with the

self-set.

Begin
while ( coverage < desiredCoverage ) do
create a GA
run GA for the defined number of generations
randomize a GA population
while (!done) do
perform crossover
perform mutation
end while
take the best detector from the population
Add the best detector into the Detector Set
Compute currentCoverage
end Begin

Figure 4.18 An evolutionary algorithm to generate multishaped detectors.