Latest Immune Models and Hybrid Approaches - Immunological Computation: Theory and Applications - page 160

license (LmatDCs), and express LmatDCs cytokines. Conversely, if the DC is
exposed to safe signals, the cell matures diff erently and becomes an unlicensed
mature DC, expressing ULmatDCs cytokines. h e LmatDCs cytokines activate
T lymphocytes expressing complimentary receptors to the presented antigen. Any
peripheral cells expressing this antigen type are removed through the activated T
lymphocyte. h e ULmatDCs cytokines suppress the activity of any matching T
cell, inducing tolerance to the presented antigen.
6.2 Multilevel Immune Learning Algorithm
A multilevel framework that combines both B- and T cell recognition mechanisms
was proposed, and called the Multilevel Immune Learning Algorithm (MILA),
which is inspired by the interaction and processes of T cell-dependent humoral
immune response (Dasgupta et al., 2003). In biological immune systems, some B
cells recognize antigens through immunoglobulin receptors on their surface, but
they are unable to proliferate and diff erentiate unless prompted by the action of
lymphokines secreted by T h cells.
Moreover, for T h cells to become stimulated to release lymphokines, they must
also recognize specifi c antigens. However, although T h cells recognize antigens
through their receptors, they can only do so in the context of MHC molecules.
Antigenic (Ag) peptides are extracted by APCs through a process similar to feature
extraction, called Ag presentation . Under certain conditions, however, B cell activa-
tion is suppressed by T suppressor (T s ) cells, but specifi c mechanisms for such a
suppression are still unknown. h e activated B and T cells migrate to the primary
follicle of the cortex in lymph nodes, where a complex interaction and kinetic pro-
cess of proliferation (cloning), mutation, selection, diff erentiation, and death of B
cells take place in germinal center chambers. h ese antibodies function as eff ectors
to the humoral response by binding to antigens and facilitating their elimination.
In MILA, an abstraction of these complex immunological events is incorpo-
rated to develop a multilevel change detection algorithm. Accordingly, the algo-
rithm consists of initialization, recognition, evolutionary, and response (as shown
in Figure 6.4).
In the initialization phase, the detection system is “trained” by giving the
knowledge of “self.” h e outcome of the initialization is used to generate sets of
detectors, analogous to the populations of T h -, T s -, and B cells, which participate in
humoral response. It is important to note that a multilevel (multiresolution) detec-
tion is considered; specifi cally, three levels of detection were introduced:
APC level, which corresponds to the highest level
B cell level, the intermediate level, is used for global pattern
T h cell, the lowest level, or bit level, for local patterns