SOME MATHEMATICAL AND COMPUTER MODELING OF NEURAL NETWORKS

The purpose of this paper. is to provide a brief background on neural networks, a summary of the mathematical models for some learning rules for neural networks, and some new computer graphics for these. The applications of these learning rules include both Boolean and continuous functions, as well as the construction of graphical mappings of the sensorary space as a two-dimensional neural grid. Numerical interpretations of the computer graphics generated far each of these learning rules are provided to serve as a guide for comparisons. The application of neural networks to solving the travelling salesman problem (TSP) is also discussed, as well as the method of simulated annealing. Computer graphics ape provided for solutions to the TSP and also for activation of an output neuron for a three-layer feed-forward network which is trained using a Boolean function. Conclusions and future directions of the research are discussed.