Neural Networks

Objectives

Learn the fundamental concepts and techniques of 'Computational Neuroscience' and of ANNs and learn to solve engineering problems with ANNs.

General characterization

Code

10992

Credits

6.0

Responsible teacher

José Barahona da Fonseca

Hours

Weekly - 4

Total - 86

Teaching language

Português

Prerequisites

It is desirable for the student to have some programming experience in Matlab but it is not required

Bibliography

1. M. T. Hagan, H. B. Demuth and M. Beale,
Neural Network Design,
1996, PWS PUBLISHING COMPANY, Boston, MA.

2. T. P. Trappenber,
Fundamentals of Computational Neuroscience,
2002, Oxford University Press, NewYork.

Teaching method

Projected slides and examples of projected computations

Evaluation method

5 Individual Practical Reports where are proposed the solution of problems that imply the profound knowledge of all the program of Neural Networks

Subject matter

1. The Natural Neuron,1.1 The Hodgkin & Huxley's Model; 1.2 The Wilson's Model, 2. Introduction to ANNs, 3. The Rosenblatt's Perceptron, 4. The Rosenblatt's Perceptron Learning Rule, 5. Limitations of One Layer Perceptrons, 6. Sigmoidal Networks, 7. The Backpropagation Algorithm, 8. Variants of Backpropagation Algorithm, 9. Introdution to Recorrent Networks. Learning in Recurrent Networks, 10. Hopfield Recurrent Network. The Hopfield Network as Associative Memory. Stability Analysis, 11. Illustrative Examples of Applications of Recurrent Neural Networks.

Programs

Programs where the course is taught:

Master in Electrical and Computer Engineering