Softcomputing

Lecturer:

Tutor:

Supervisor:

Department of Cybernetics

Synopsis:

The aim of this course is to get the students knowledgeable with non-traditional computational techniques of optimisation, state-space search, control and decision-making. Many of the softcomputing methods utilise analogies with various phenomena in nature and/or society. Results obtained by these methods often have a high quality, but their absolute reliability is never guaranteed. During the seminars the students will get a chance to get basic practical skills with a sample softcomputing problem.

Requirements:

For successful completion of the course, it is necessary to present the results of the individual work to other students and explain the approaches used.

Syllabus of lectures:

1. Introduction to softcomputing methods, relationship to phenomena known from other scientific fields

2. Evolutionary algorithms - Introduction to EAs, history and survey of the EA techniques, simple GA, applications of EA.

3. Evolutionary algorithms - Schema theory, example of GA, genetic programming, grammatical evolution, example applications.

4. Evolutionary algorithms - Strongly typed GP and grammatical evolution, application examples. Decision tree induction using GP.

5. Evolutionary algorithms - Multicriterial optimization.

6. Evolutionary algorithms - Competent genetic algorithms. Premature convergence and how to prevent it.

7. Estimation of Distribution Algorithms - EDAs for discrete representations.

8. Estimation of Distribution Algorithms - EDAs for real representations.

9. Neural networks - Introduction to neural networks, model of a simple neuron and its similarities with linear regression. Multilayer feedforward neural networks, backpropagation.

10. Neural networks - Limitations of backpropagation, learning the structure and weights of NN, cascade-correlation NN, NEAT, NERO.

11. Neural networks - Hopfield's net, Kohonen's self-organizing map, unsupervised learning, radial neuron, analogy with cluster analysis, extension for classification, initial weight estimate, application example.

12. Fuzzy logic and decision making.

13. Fuzzy modelling and control.

14. Summary (spare space)

Syllabus of tutorials:

1. Organisational matters, seminars/labs detailed contents

2. Evolutionary algorithms - simple GA example, assignments on EAs

3. Evolutionary algorithms - individual work on assignments

4. Evolutionary algorithms - examples of modified GAs with limited convergence, individual work on assignments

5. Evolutionary algorithms - individual work on assignments

6. Evolutionary algorithms - applications of EAs, individual work on assignments

7. Evolutionary algorithms - individual work on assignments

8. Evolutionary algorithms - competent EAs, searching interactions among variables

9. Neural networks - introduction to program equipment for NN creation and application, example of NN training and its application to a typical task

10. Neural networks - assignment: classification of objects to 2 classes using multilayer perceptron.

11. Neural networks - finishing and hand in of the second assignment.

12. Fuzzy systems - examples

13. Test

14. Summary, (spare space)

Study Objective:

Study materials:

There is no text-book covering the course completely; any book on modern operating systems can be used. The lecturer will hint resources to particular topics.

Note:

Further information:

No time-table has been prepared for this course

The course is a part of the following study plans:

• Cybernetics and Measurements - Control Engineering- structured studies (compulsory elective course)
• Cybernetics and Measurements - Artificial Intelligence- structured studies (compulsory elective course)
• Cybernetics and Measurements - Measurement and Instrumentation Systems- structured studies (compulsory elective course)
• Cybernetics and Measurements - Aeronautical Engineering and Control Systems- structured studies (compulsory elective course)