Systems Engineering Neural Networks

5Practice Makes Perfect

[…]The perception and definition of a particular system, its architecture and its system elements depend on an observer's interests and responsibilities.[…]

ISO/IEC 15288

In this chapter nine interesting examples will be presented, some developed with the MS Excel^® spreadsheet and others with Python programming language. These exercises can be useful for not only understanding the algorithms based on mathematical models³⁰ for decision-making but also to provide a guideline for other applications for various needs. Therefore, it will be shown that neural networks are useful for problem solving, processing approximate data and providing consistent results of the examined event. Furthermore, we will see that neural networks can learn with a certain level of uncertainty, which can be already specified in the input data set.

The supervised learnings will be explained when the neural network solves a mathematical function, characterized by stationary points (let us only conform to the minima and maxima). Furthermore, such functions will allow us to reach or correctly set the optimum point of our model. If we believe that neural networks can provide a significant advantage when the cost function is introduced, it is important to explain that the cost function is an essential element for the optimization process in Linear Programming (LP). Considering that one of the main managers' objective is to allocate resources among activities or projects. LP is a widely ...