Search

Nowadays artificial neural networks (ANN) are quite a popular computational tool, considered as a form of artificial intelligence. The paper presents the structure of artificial neuron and functioning of those types of neural networks that are most often applied to solve medical problems (feedforward and recurrent networks). Training of ANN -- supervised and unsupervised -- was also described since it is an essential stage of building all neural network models. Finally, the tasks performed by neural networks were characterisedand divided in 2 categories: classification and regression problems. The second part of text contains the review of medical applications of ANN with, which was prepared with special attention given to endocrinological and diabetological issues as well as to research done by Polish authors. [ABSTRACT FROM AUTHOR]

This paper presents neural networks in process quality control. The control is a very important step in the production process. There are different methods for checking the quality of the process. In some manufacturing companies quality is checked by machine operators or quality control departments. Here can be used method for manually checking the quality or tools of statistical control in the form of specialized computer programs. Another approach is to treat the control card as an image instead of a set of points. We have developed the models of analysis of process control charts. The models were created in the form of multilayer networks with error back propagation (MLP neural networks). The models were tested on real data from the production company. [ABSTRACT FROM AUTHOR]