ECE 542 Neural Networks
 

Introduction to neural networks and other basic machine learning methods including radial basis functions, kernel methods, support vector machines. The course introduces regularization theory and principle component analysis. The relationships to filtering, pattern recognition and estimation theory are emphasized. 3 credit hours.

 
   

• Prerequisite
 

ECE301 (Signals and Systems), ST371 or MA421 (Probability and Statistics), The course will use elementary matrix algebra, includes matrix/vector multiplication, matrix inverses, matrix description of the solution to linear equations, eigenvalues and eigenvectors. This material is covered in ECE220. The course will heavily utilize MATLAB for homework and examples in class. MATLAB is required for ECE220 and many other courses at NCSU. The course will use elementary probability theory: means, variance, covariance, expected values/moments of linear functions, stationarity.


• Course Objectives
 

Upon completion of the course, the student will be able to

  • Describe appropriate applications of  neural networks and learning methods
  • Compare neural network solutions to other methods of modeling, approximation, simulation, classification, pattern recognition and decision making
  • Determine which methods of learning are appropriate for a given problem
  • Formulate and design neural networks and other learning systems to solve realistic applications
  • Measure the performance of the neural networks and other learning methods

• Course Requirements
 

Assignment

Weight

Homework (~7)

30%

Midterm exam

35%

Final exam

35%

Homework will be submitted via Wolfware in pdf format. The student should have a scanner available to scan any handwritten solutions.


• Textbook
 

Neural Networks and Learning Machines, Simon Haykin, Pearson/Prentice-Hall, 2008, ISBN 978-0-13-603219-9
Please review sections 4.4, 5.2-4, 6.2-3 of the text to assure that your background is sufficient.


• Computer and Internet Requirements
 

NCSU and Engineering Online have recommended minimum specifications for computers. For details, click here.

MATLAB is available for free at download.eos.ncsu.edu to students registered for at least one NCSU engineering class. (Unity ID required)

Students must carefully follow instructions for installing MATLAB, and need to download Virtual CloneDrive only if they are using Windows.

When trying to access MATLAB or other software from off campus you’ll need to use VPN. If you have problems with the download or installation process, contact the Help Desk located in Daniels Hall at eoshelp@ncsu.edu.

Another option for accessing software such as MATLAB, SolidWorks and several other engineering applications is through the Virtual Computing Lab (VCL) at: http://vcl.ncsu.edu/.


• Instructor
  Dr. Joel Trussell, Professor
Electrical & Computer Engineering
Engineering Building II (Eb2) 2058, Box 7911
Raleigh, NC 27695

Phone: 919-515-5126
Fax: 919-515-5523
Email: hjt@ncsu.edu
Website: http://www.ece.ncsu.edu/people/hjt