Knowledge Discovery and Data Mining in Medicine - Advanced Techniques in Knowledge Discovery and Data Mining - page 192

allocation module to distribute a training case and some classification modules to
classify a set of training cases. Each classification module consists of a worker
neural network (worker NN) and a monitor neural network (monitor NN). The
monitor NN estimates how conformable the worker NN is to a given training case.
The training cases are distributed over different classification modules iteratively
according to the learning condition of a training case, until the sum of the squared
errors reaches an expected value. We consider that each classification module
competes with the others in the classification of training cases. Reflective NN has
an outstanding classification capability, even if there are missing data or
information, as is common in medical databases.
However, the optimal number of classification modules in reflective NNs is
not determined according to the probability distribution function in the space of
training cases, even if we have approached finding the optimal structure with
structure level adaptation of the neural network [7]. To solve such a problem, we
find some relation between the number of modules and the number of subsets of
training cases by our proposed classification method of immune multiagent neural
networks (IMANN) [1]. The IMANN has macrophage agents, T-cell agents, and
B-cell agents. The macrophage and T-cell employ the planar lattice neural
networks (PLNN) with the neuron generation/annihilation algorithm [2]. This
network structure consists of hidden neurons in the lattice. The network can work
in a manner similar to self-organized map (SOM) [3]. B-cell employs Darwinian
neural networks (DNN) [4], which have a structural learning algorithm based on
Darwin's theory of evolution.
Fig.7.9. Reflective NNs.