The Methodical Complex of Laboratory Works on the Study of Neural Network Technologies

The Methodical Complex of Laboratory Works on the Study of Neural Network Technologies

Artem Borodkin (National Research University “MPEI”, Russia), Vladimir Eliseev (National Research University “MPEI”, Russia), Gennady Filaretov (National Research University “MPEI”, Russia) and Alireza Aghvami Seyed (Payame Noor University, Iran)
Copyright: © 2019 |Pages: 10
DOI: 10.4018/978-1-5225-3395-5.ch030

Abstract

The chapter considers a task of teaching undergraduate students practical skills using artificial neural networks to solve problems of information processing and control systems. It represents and proves the methods of teaching, based on the gradual increase in the complexity of tasks to be solved by students. The developed complex of laboratory works includes classical problems and methods of their solutions, as well as original methods for solving problems of automatic control. The technology base of the laboratory works are both well-known programs and software package developed by the authors. In addition to the practical experience in the use of software packages, students obtain experience in conducting comparative studies of traditional and neural network methods for solving control problems.
Chapter Preview
Top

Background

As a result of the analysis of experience and the available sources we can distinguish three levels of presentation of the information about neural networks in depth and detail:

  • 1.

    as a part of the course on intellectual technologies;

  • 2.

    as a separate course involving the consideration of different neural network algorithms, but without specialization on the applications;

  • 3.

    as a special course which culminates in a discussion of examples of neural networks application for solving various problems.

It should be noted that the courses of the first type because of its breadth does not allow students a deep understanding of the neural networks. Considering different Intelligent technologies, including expert systems, fuzzy logic and genetic algorithms, duration of one semester course is usually enough only for exposition of the basic principles and the most common neural network paradigms (Makarenko, 2009). If such courses include laboratory work, that is only for acquaintance with any single neural network paradigm on a typical example (Benjaminsson, 2008).

Key Terms in this Chapter

NNC: Neural network controller – a control algorithm which is based on ANN principles and implementation.

Backward Propagation of Errors: A gradient descent algorithm to train artificial neural network.

Automatic Control Systems: A class of technical systems with automated way of processing information to reach some goal defined over measured values in terms of some kind of quality.

Educational Software: A set of programs used to facilitate knowledge adoption by pupils.

Methodical Complex: A set of principles and practical considerations to provide steep learning curve.

CDIO: Conceive, design, implement, operate – an education framework that blends theory.

Complete Chapter List

Search this Book:
Reset