B Cell–Inspired Algorithms - Immunological Computation: Theory and Applications - page 148

Jerne
Varela
Perelson

Farmer

Hunt and Cooke

Other models aiNet

Tshiguro AISEC AINE

aiNet
hierarchy

IPD aiNet

opt-aiNet

Michelan and
Von Zuben

Reactive
IN

RLAIS

Fuzzy AIS SSAIS

CLARINET

TECNO-STREAMS Metastable

IN

Fractal
IN

Figure 5.15 Chronological tree of AIN models. RLAIS , Resource limited
artifi cial immune system; AISEC , Artifi cial immune system for e-mail
classifi cation. (From González, F., J. Galeano, and A. Veloza in Proceedings
of the 2005 Conference on Genetic and Evolutionary Computation (GECCO'05),
ACM Press, Washington, 2005, 361-368.)

algorithm is not the B cell but the ARB or cluster. Another model called aiNet,
which shares some characteristics of Timmis' AINE, was proposed (De Castro and
Von Zuben, 2001). h is work emphasizes its self-organizing ability, that is, the use
of minimal number of control parameters; several versions of AiNet with diff er-
ent enhancement are subsequently developed (De Castro and Timmis 2002a,b,c;
De Castro and Von Zuben, 2002a,b; De Castro, 2003). Timmis and Edmonds (2004)
have done further analysis of Opt-AiNet and commented on its implementation.
Nasraoui et al. (2002, 2003a) introduced an IN-based algorithm (called
FuzzyAIS) that uses a fuzzy set to model the area of infl uence of each B cell. h is
improves the expression of earlier models and makes it more robust to noise and
outliers.
In their recent work, González et al. (2005) developed a general model of AiNet
(Figure 5.15), which provides a common notation and description of AiNet and its
variation. Many applications of immune network models are reported in the litera-
ture (Ishiguro et al., 1994, 1996; Luh and Liu, 2004; Michelan and Zuben, 2002;
Mitsumoto et al., 1996; Secker et al., 2003; Timmis and Edmonds, 2004; Timmis
et al., 2004; Timmis and Neal, 2001).