Abstract:
The Negative Selection Algorithm developed by Forrest et al. was inspired by the way in which T-cell
lymphocytes mature within the thymus before being released into the blood system. The mature T-cell
lymphocytes exhibit an interesting characteristic, in that they are only activated by non-self cells that
invade the human body. The Negative Selection Algorithm utilises an affinity matching function to
ascertain whether the affinity between a newly generated (NSA) T-cell lymphocyte and a self-cell is less
than a particular threshold; that is, whether the T-cell lymphocyte is activated by the self-cell. T-cell
lymphocytes not activated by self-sells become mature T-cell lymphocytes. A new affinity matching
function termed the feature-detection rule is introduced in this paper. The feature-detection rule utilises
the interrelationship between both adjacent and non-adjacent features of a particular problem domain to
determine whether an antigen is activated by an artificial lymphocyte. The performance of the featuredetection
rule is contrasted with traditional affinity matching functions, currently employed within
Negative Selection Algorithms, most notably the r-chunks rule (which subsumes the r-contiguous bits
rule) and the hamming distance rule. This paper shows that the feature-detection rule greatly improves
the detection rates and false alarm rates exhibited by the NSA (utilising the r-chunks and hamming
distance rule) in addition to refuting the way in which permutation masks are currently being applied
in artificial immune systems.
