| Abstract [eng] |
The paper describes what to consider when constructing multi-classifier systems (MCS), what is perceptron, neural network (NN), how to create NN. We analyzed the work of other authors when they combined NN and used perceptron as combination function. Various types of theoretically optimal weights calculation (TOWC) functions are analyzed. Finished experiments. The aim of this work is to investigate whether it is possible to improve the quality of MCS using the initial TOWC as initialization of weights of combiner-perceptron. The carried out research shows that we can improve the quality of MCS, but the combinator-perceptron initialized by random weights will more often return a better result. |
