General information about the artificial neural net Neuronix
The idea of artificial neural networks (ANN) is to simulate some features of brain. ANN is the only very siplified model of some of the functionality and elements of the brain, mainly neurons and synapses. There are different architectures of ANN as well as different metods of learning by them. The knowledge necessary to solve problems is acquired at the stage of learning. Neuronix uses method of supervised learning with backpropagation and during this process the networks is to be repeatedly presented example input patterns with o target or desired output pattern (in other wors – correct answer). In Neuronix thanks to the wizards the process of learning supervision is quite easy. The process of learning is autmatically finished after achieving some degree of satisfying error.
Potential areas of application of Neuronix are:
- estimation of credit risk,
- forecasting of financial results,
- sales forecasting,
- forecasting the stock market,
- pattern recognition, the handwriting,
- data analysis,
- objects classification,
- analysis of the acoustic signals,
- signals noise filtering,
- and many others.
Very important features of the Neuronix are:
- its own translator of the same knowledge representation and procedural language as in the expert system shell – PC-Shell,
- possibility of building fully integrated hybrid applications combining Neuronix applications with that of the PC-Shell.