The strength of Artificial Neural Networks (ANNs) derives from their perceived capability to infer complex, non-linear, underlying relationships without any a priori knowledge of the model. However, in reality, the fuller the priori knowledge the better the end result is likely to be. In this paper we show how to implement the Gamma test. This is a non-linear modelling and analysis tool, which allows us to examine the input/output relationship in a numerical data-set. Since its conception in 1997 there has been a wealth of publications on Gamma test theory and its applications. The principle aim of this paper is to show the reader how to turn the Gamma test theory into a practical implementation through worked examples and a explicit discussion of all the required algorithms. Furthermore, we show how to implement additional analytical tools and articulate how to use them in conjunction with the non-linear modelling technique employed.
|Number of pages||9|
|Journal||International Journal of Simulation: Systems, Science and Technology|
|Publication status||Published - 1 Jan 2005|
- Near neighbours
- Noise estimation
- Non-linear modelling