**Welcome to**

*Neural and evolutionary computing is an emerging technology in advanced
computer-applications. It is about using computers to learn, optimise and discover new
engineering solutions and technologies by mimicking human intelligence in the brain and
genetic systems. It has been proven very useful in learning and automating (as well as
discovering new) designs for electronic/electrical systems, circuits, filters, motors,
drives and control systems. This course however requires little mathematics or
programming.*

Press here for an interactive answer and hands-on excercise.

- Executable DOS file developed by an NEC4'95 student and the Lecturer: nec4tsp.exe
- Source code in C (Travelling Salesman's Problem and Simulated Annealing): nec4tsp.cpp
- Borland C++ IDE (integrated development environment) file: nec4tsp.ide

- Thanks to University of Michigan and Carnegie Mellon University, you can now carry out "distance-learning" and self-tutorials of MATLAB by following one of these links:

http://hpme12.me.cmu.edu/Matlab/html/

http://www.engin.umich.edu/group/ctm

With access to both the WWW and to Matlab on either a personal computer or a workstation, any student will be able to follow a tutorial while running Matlab and will be able to easily switch between the two programs. Students will be able to grasp key concepts and design techniques in a "learn by seeing and doing" manner.

Commands shown in the tutorials can be copied from the Netscape, MSIE(?) or Mosaic window and pasted into the Matlab command window or into an m-file with a simple point and click of the mouse; there is no need for time-consuming typing and editing. Students can immediately see the result of an actual computation, compare it to the result shown on the tutorial, and quickly experiment with modifications of the commands and changes of parameters.

- You can escape from the above just now, and only follow the instructions and simple
MATLAB code provided in the Laboratory Sheets. Also try to follow the on-line help in the
MATLAB window and consult the "command sheets" and manuals provided in the
Laboratory. If you wish to download (and not to type in) the m-file used for the 3-D and
its contour plots,
**multipk.m**, click here and save it in your**c:\matlab\bin**directory.

Follow the instructions given in the Laboratory Sheets.

- FlexTool(GA) Toolbox
- Exhaustive search and Optimisation
- 1-D Single Objective Single-Peak and Multi-Peak Problems
- 2-D and Multiple Objectives

Follow the instructions given in the Laboratory Sheets.

Backpropagation Learning

Application in Noise Cancellation

Hebbian Learning

Solving Neural Learning Problems in Tutorial 2

(Optional and during spare time only)

The IEE Video Tape shown at the lecture can be viewed again from the CD-ROM: "IEE Distance Learning - Neural Computing Video Course (SO2). You can run this on the Pentium 133 PCs, which have a sound card. Ask Mr. O'Hara, the Technician, for a pair of speakers. You can also use the CD to see the demonstration on how neural networks work. Note that other options do not work on the CD!

You can also borrow the IEEE/ABAS "Learning Neural Networks" CD-ROM from Mr. O'Hara.

"1993 IEEE International Conference on Neural Networks" CD-ROM also available.

"Second IEEE International Conference on Fuzzy Systems" CD-ROM also available.

This page must not be used alone and must be used in conjunction with NEC4 Lecture Notes, Appendices, Tutorial Sheets, Laboratory Sheets and Assignment Sheets. The IEE and IEEE video tape and CD-ROMs used in the course may be borrowed from the Lecturer, Dr. Y. Li.

Syllabus is here. Sister Course to merge later: Image Processing and Pattern Recognition IV by Prof John Barker.

Authors |
Title, edition |
Publisher |
Year |
ISBN |
Cost |
Code |

Z Michalewicz | Genetic Algorithms + Data Structures = Evolution Programs | Springer-Verlag, 2nd Ed | 1994 | 3540580905 | £25 | B |

S V Kartalopoulos | Understanding Neural Networks and Fuzzy Logic | IEEE Press | 1996 | 0780311280 | $35 | B |

David E Goldberg | Genetic Algorithms in Search, Optimisation and Machine Learning | Addison-Wesley | 1989 | 0201157675 | £24 | C |

Codes : A = compulsory; B = strongly recommended; C = recommended; D = wider reading

- See Laboratory Sheets for MATLAB code that may be revised for solving Q.1, Q.4 and Q.5 using FlexTool(GA) toolbox.

- Q.3 MATLAB code: T1Q3Training.m, T1Q3Plotting.m
- Q.5 MATLAB code: T1Q5

- Start to
*think about it now*! You are also highly encouraged to propose your own mini-project.

Completed assignment must be submitted to Dr. Li (into a
"run-about" mailbox just outside B1-16) by the end of **Week 12, Friday, 23
January 1998**. If you have genuine difficulties beyond your control, however,
you may be allowed to submit the entire mini-project by Week 13, but you must contact Dr. Li in the first instance to submit an outline
plan of your mini-project by Week 11. Submitting a floppy containing your programs is
voluntary, but you may be asked to do so for random inspection.

If your report is of value for future teaching, we would like to ask for your contribution to this course the next year.

*
Which
algorithm might be the most appropriate to your application at hand:*NB. Swarm
intelligence such as particle swarm optimisation and ant colony optimisation
fits in the 'evolutionary programming' - 'evolution strategies' branch for
continuous or numerical optimisation and fits in the 'genetic algorithm' branch
for discrete or structural optimisation.

The following
classification illustrates the complexity of computational problems, where **
NP-complete** are a class of computational problems that cannot be solved in
deterministic polynomial (**P**) time but can be solved in nondeterministic
polynomial (**NP**) time. Given the modern computer simulation power,
many virtual engineering problems can now be solved via digital prototyping,
although in exponential time (i.e., they may be *theoretically solvable but
practically intractable*). **
The power of evolutionary computation lies in its ability to solve many
exponential problems in NP time, i.e., to make exponential problems practically
solvable.** Some problems such as winning the lottery, however,
remain an exponential problem (i.e., is 'exponential-complete' and can only be
solved by enumerations).

All materials provided by this site are copyright protected and are for use with Glasgow University's *Neural and
Evolutionary Computing IV* (1995-97) course only. It is illegal to copy, use or
distribute any such material for any other purposes without the consent of the course
Lecturer, Dr. Y. Li.