NEURAL NETWORKS

Based on a set of notes prepared by Dr. Patrick H. Corr (Computer Science)

What and why?

Neural Networks: a bottom-up attempt to model the functionality of the brain.

Two main areas of activity:

Biological
- Try to model biological neural systems
Computational
- Artificial neural networks are biologically inspired butnot necessarily biologically plausible
- So may use other terms: Connectionism, Parallel Distributed Processing, Adaptive Systems Theory.

Symbol	Subsymbol

top-down	bottom-up
explicit	implicit
rules	examples
serial	parallel
digital/boolean (true or false)	analog/fuzzy
brittle	robust

Interests in neural network differ according to profession.

Neurobiologists and psychologists: understanding our brain
Engineers and physicists: a tool to recognise patterns in noisy data (see Ts at right)
Business analysts and engineers: a tool for modelling data
Computer scientists and mathematicians: networks offer an alternative model of computing: machines that may be taught rather than programmed
Artificial Intelligensia, cognitive scientists and philosophers: Subsymbolic processing (reasoning with patterns, not symbols)

Now go on to read more in detail about how Neural Networks are built. Follow these links in turn to my notes which attempt a qualitative explanation.

Biological Foundations
An Shockwave Flash animation of an Artificial Neural Network being trained to recognise flags.
Techniques (The Multi-Layer Perceptron)

For those who want a more rigorous, mathematical explanation, have a look at one of the following sets of notes on the WWW.

An Introduction to Neural Networks (Stirling) One long page, covering everything at a gallop.
Neural Nets by Kevin Gurney. A much longer, more detailed explanation, from which only the introduction and maybe the sections on errors, back-propagation and applications are of interest to us.

Then take a look at:

Input -> Output	Model:	of human expertise	learned from data
Facts -> Decision	Logic representation (facts and rules)	Expert system	Machine induction
Numbers (measurements -> predictions)	Mathematical calculation	Decision support system	Neural network

Prepared by Dr. David R. Newman from Pat Corr's notes.