Question: Apply the aiNet algorithm to the data set presented (Exercise). Make use of the MST edge inconsistency criterion in order to separate the network clusters and thus identify the clusters of the original data set. Use the same parameters as those used to solve the SPIR problem, including the factor r = 2. Compare the performance of both algorithms: ACA and aiNet.

Exercise: Given the data set illustrated in Figure, use the negative selection algorithm to generate a set of N = 1,000 detectors that recognize everything but the self patterns. Assume a cross-reactivity threshold ε = 72, corresponding to 60% of the length of each pattern. The affinity measure is given by Affinity = L - D, where L is the length of the strings (L = 120), and D is the Hamming distance between two strings (Equation).

Evaluate the set of detectors generated by randomly introducing noise in the samples, with various noise levels from 5% to 50%, and monitoring for the presence of unknown patterns. (Patterns with more than 40% of noise should be detected as nonself.)