T Cell–Inspired Algorithms - Immunological Computation: Theory and Applications - page 110

It can hide the self-concept (Esponda et al., 2004). At the other level, it has
various strengths that may not be totally unique to this method.
Compared with other change detection methods, NSAs do not depend on
the knowledge of defi ned “normal.” Consequently, checking activity of each
site can be based on a unique signature of each while the same algorithm is
used over multiple sites.
h e quality of the check can be traded off against the cost of performing a
check (Forrest et al., 1994).
If the process of generating detectors is costly, it can be distributed to mul-
tiple sites because of its inherent parallel characteristics.
Detection is tunable to balance between coverage (matching probability) and
the number of detectors (D'haeseleer, 1996).
Some limitations of the string representation in NSA are as follows:
Binary matching rules are not able to capture the semantics of some complex
self/nonself spaces.
It is not easy to extract meaningful domain knowledge.
In some cases, a large number of detectors are needed to guarantee a desired
level of detection (Scalability issue).
It is di cult to integrate the NS algorithm with other immune algorithms.
Crisp boundary of self and nonself may be hard to defi ne.
In real-valued representation, detectors are represented by hypershapes in an
n -dimensional space. h e algorithms use geometrical heuristics to distribute the
detectors in a uniform way on the nonself space.
Some limitations of the real-valued representation in NSA are as follows:
h e issue of geometrical shapes
h e handling of dimensionality
h e estimation of coverage
h e selection of distance measure
Ji and Dasgupta (2006), in a recent study, pointed out that NSAs are not appro-
priate to be used as a general classifi cation method because they use samples from
one class in training. h ere exist inconsistency in the terminology and the assump-
tions used in NSAs to clarify the confusion or misunderstanding in applicability
(Freitas and Timmis, 2003). Compared with binary or string representation, for-
mal analysis is needed for the genre of real-valued representation.
NSAs play an important role in the research of artifi cial immune systems (AIS).
Chronological development of NSA and its variants are reported by Ji and Dasgupta
(2007).