ECE: Electrical & Computer Engineering

ECE 5724 Neural and Fuzzy Systems


Fall 2014 textbook list

The Fall 2014 ECE textbook list is available online for students.

Current Prerequisites & Course Offering

For current prerequisites for a particular course, and to view course offerings for a particular semester, see the Virginia Tech Course Timetables.

Return to course list

ECE 5724 Neural and Fuzzy Systems (3C)

Introduction to various structures of artificial neural networks and fuzzy logic systems, as well as special learning mechanisms such as generalized back-propagation, clustering and genetic algorithms. Applications will be made to classification problems, binary associative memories, self-organizing maps, and nonlinear system modeling and control including on-line adaptation.

What is the reason for this course?

This course fills a needed gap in the general (standard) techniques of system analysis and design which are currently taught in the curriculum. For example, since this course uses “nonparametric” methods, there is no requirement for the process to be linear in order for the material in this course to be relevant.

Typically offered: Spring. Program Area: Systems/Controls.

Prerequisites: Prerequisites: 5704.

Why are these prerequisites or corequisites required?

The prerequisite of ECE 5704 is to ensure that the student obtain the proper sophistication in the controls/systems area. This material includes state space models and the techniques of system identification/digital control. The necessary topics are also found in most DSP classes.

Department Syllabus Information:

Major Measurable Learning Objectives:
  • Model the neuron and its components and apply various learning rules to problems.
  • Apply various kinds of neural networks and the back propagation algorithm to problems of interest.
  • Apply fuzzy sets and fuzzy rules for solving control problems.
  • Apply a number of iterative learning algorithms which solve problems of identification and control of systems based only on available input-output data, as well as many other classical problems of interpolation and classification.

Course Topics
Topic Percentage
Introduction:Machine learning, classification and control 5%
Introduction to Artificial Neural Nets (ANNs) 30%
Artificial neurons
One-layer perceptrons/Adalines 3%
Correlation matrix memory/BAMs 3%
Self-organizing maps (SOMs) 6%
Clustering/radial-basis function (RBF) nets 6%
Multi-layer perceptrons (MLPs) 12%
Data scaling and preprocessing
Backpropagation training
ANNs for System Identification 15%
Nonlinear mapping and function approximation
Identifying a chaotic generator 3%
System modeling: ARMA models 3%
Linear dynamic system identification 3%
Dynamic neural nets - a canonical structure 3%
Nonlinear dynamic system identification 3%
Introduction to Fuzzy Logic 22%
Linguistic variables/rules 4%
Fuzzy sets and relations 6%
Minimum and product inference 6%
Fuzzy logic system (FLS) identification 6%
Genetic algorithms (GAs)/classifier systems 8%
Applications to system control 20%
Inverse plant modeling 5%
Disturbance rejection 5%
Model predictive control 5%
Adaptive control 5%

Return to course list