"Artificial Neural Nets and Problem Decomposition"

Supervised by A.Sharkey

Abstract

Since the publication of Rumelhart and McClelland's seminal work on neural network theory, in 1986, the field has grown and grown as people come to realise the potential that neural networks have for dealing with complex pattern recognition problems. The ability to model highly non-linear, non-parameterised data in ways previously impossible using more traditional, mathematical methods, means that neural nets are being used in applications that though previously to be impossible to model computationally.

The stock market is the most glamorous and, at the same time, the least understood of the financial markets, and for years and years people have tried to forecast the future behaviour of the market, often with disastrous consequences. The stock market is so complex in its exact function that it is considered by many to be more or less random, although in recent years people have been identifying non-linear patterns within the activity of the market.

Actually modelling the stock markets therefore, seems like the sort of problem where a neural net may well succeed where others have failed. Various kinds of network are used to try and achieve this, and the results provide useful insight into the abilities of networks when dealing with this sort of problem.