Knowledge Discovery and Data Mining in Medicine - Advanced Techniques in Knowledge Discovery and Data Mining - page 193

Fig.7.10. A model of the biological immune system.

7.4.1 Biological Immune System

The living body maintains normal condition by its biological immune system,
where various immune cells perform their own functions and cooperate with each
other. The biological immune system mainly works to protect a body from many
antigens. Immune cells learn to recognize relevant patterns, they remember the
patterns that have been seen previously, they fight with antigens using their
patterns, and they promote robustness for unknown patterns by using their
diversity. Figure 7.10 shows the relationship of immune cells.
The basic idea of our proposed IMANN is a cooperation and competition
mechanism in a biological immune system. The macrophage can recognize
unknown patterns that invade its living body at first. A T-cell receives the
stimulation by macrophages and works to promote the B-cell's activities. And
then B-cells manufacture their specified antibody. The IMANN is developed to
imitate the characteristic functions of a biological immune system.

7.4.2 Planar Lattice Neural Networks

PLNN is a type of three-layered neural network, but the neurons in the hidden
layer are arranged on a lattice. The network can work similar to SOM, that is, the
patterns between inputs and outputs are classified into some groups in the lattice.
Moreover, we can expect to extract some If-Then rules from the trained network.
Figure 7.11 shows an overview of PLNNs [2]. The neural network is a kind
of layered neural network, consisting of an input layer, an interconnected hidden
layer, and an output layer. The interconnected hidden neurons adjust the
connection weights between input neurons and hidden neurons according to the
relation of input-output patterns and the neighborhood N of the hidden neuron i .
N is a set of neighboring neurons around i on the lattice. If S is the set of
neurons and

} {

N = N i ∈ S is the set of neuron neighborhoods, then the pair

i

{ }

S , forms a graph in the usual sense. The neighborhood system shown in Fig.
7.12 belongs to the class of homogeneous neighborhood system (HNS) defined as
follows.

N