EE 456 – Introduction to Neural Networks

Designation:

Senior/Grad-level technical elective for Electrical Engineering students

Catalog Data:

Solving problems with artificial neural networks in pattern classification and function approximation, application of feedforward and feedback neural networks. Prerequisite: CMPSC 201C or CMPSC 201F, MATH 220.

Prerequisites by topic:

1. Understanding the basics of numerical methods to solve engineering problems.
2. Understanding the basics of development and implementation of algorithms.>/li>
3. Understanding the solution of a system of linear equations and matrix algebra.

Course Objectives:

This course provides introduction to artificial neural networks as a tool for solving difficult problems for which conventional methods are not effective. Through lecture, out-of-class assignments and a group semester project, students are provided learning experience that enables them to:

1. Become proficient with computer skills (e.g., MATLAB) for the basic analysis and design of neural networks.
2. Develop technical writing and oral presentation skills important for effective communication.
3. Acquire teamwork skills for working effectively in groups on a project.
4. Identify a problem and provide a neural network based solution.

Topics:

1. Introduction (Ch. 1, 1 week)
2. Simple neural nets for pattern classification (Ch. 2, 4 weeks)
3. Backpropagation neural net and radial basis functions (Ch. 6, 4 weeks)
4. Pattern association (Ch. 3, 1 week)
5. Neural nets based on competition (Ch. 4, 1 week)
6. Other neural networks (Ch.7, 1 week)
7. Applications of neural networks (includes project presentations 1 week)
8. Two in-class examinations and one quiz (1.5 weeks)

Class Schedule:

Two 75-minute lectures per week

Computer Usage:

MATLAB 6.0 and Neural Network Toolbox were used for homework and semester projects. To facilitate ready access, the software resources were made available in the EE Computer Lab and the Penn State Network.

Laboratory projects and/or assignements:

1. Five sets of out-of-class assignments were given. Some of these assignments were submitted electronically and the remaining were hand delivered for grading and feedback.
2. Project teams comprised of one to three students per team were formed at the discretion of the students. Each team prepared a proposal that was approved after review and appropriate feedback. An interim progress report was followed by a final report and in-class presentation.

Contribution to meeting the professional component:

This course provides a design emphasis, basic analysis skill and technical writing as well as oral presentation capabilities. Topics pertaining to problem solving and offering of neurocomputing solutions in a variety of applications were considered.

Relationship to program outcomes:

The course relates to the following program outcomes:

1. Graduates will understand how to analyze and design simple artificial neural networks. [Ref: Outcome O.3.1 and Outcome O.3.2.]
2. Graduates will have attained computer proficiency. [Ref: Outcome O.1.3.]
3. Graduates will have teamwork skills. [Ref: Outcome O.5.1.]
4. Graduates will possess oral and written communication skills. [Ref: Outcome O.5.2.]