Latest Immune Models and Hybrid Approaches - Immunological Computation: Theory and Applications - page 162

h e operators used to combine diff erent pattern recognition mechanisms can be
chosen in diff erent ways either to achieve more fault-tolerant recognition or sen-
sitivity to small changes. MILA used an rcb-like matching rule in a real-valued
representation, which may be thought of as a partial euclidean distance. A T h cell
uses a slide window to get the w elements. However, a B cell uses randomly chosen
w elements. h e concept of permutation mask and crossover closure from string
presentation can be used the way these w elements are chosen for B cells.
Another feature of MILA is the implementation of positive selection by the so-
called T s cells (positive detectors), which are based on self-samples. An evolutionary
phase in MILA is a process of refi ning the detector set if the earlier detection rates
can be evaluated (verifi ed). h is phase involves cloning, mutation, and selection;
however, cloning in MILA is a targeted one (not blind), only those detectors that
are activated in the recognition phase can be cloned.

6.3 Major Histocompatibility Complex-Based Systems

To detect virus-infected cells that have been damaged internally, MHC molecules
provide a mechanism to see what happened inside a cell because under normal
circumstances, one cell cannot look inside another. h e primary role of MHC is to
display antigen fragments to T cells, particularly, MHC class I binds to CD8 on
cytotoxic (or killer) T cells, whereas MHC class-II proteins present antigen frag-
ments on an APC surface for binding to the CD4 receptor on the T h cells. By
allowing T cells to examine the internal state of other cells, the MHC mechanism
acts as a kind of cell-level anomaly detector that allows the immune system to
uncover virus-infected cells. A similar analogy of cells with running programs,
and MHC peptides with short sequences of system calls was fi rst proposed in 1996
(Forrest et al., 1996), which could detect malicious “running” programs such as
viruses and worms.
All biological systems maintain a stable internal state by monitoring and
responding to internal and external changes. h is self-monitoring is one of the
defi ning properties of life a nd is k nown a s “ homeosta sis.” Homeostatic mecha nisms
are usually autonomic, and the purpose is to minimize variations in the internal
state of an organism. Somayaji and Forrest (2000) developed process homeostasis
(pH), a Linux kernel extension that delays the execution of unusually behaving
processes. h e delay mechanism of pH was a predecessor to the network-level virus
throttling (Twycross and Williamson, 2003), where the rate of network connec-
tions was throttled to limit the spread of malicious programs (Somayazi, 2007).

6.4 Cytokine Network Model
Various immune cells in the biological immune system mutually infl uence one anoth-
er's activities through hormonelike intercellular messenger molecules called cytokines.
h e cytokines produced by one cell can modulate the production and secretion of