About the Impact of Informative Features Selection in the Mutually Orthogonal Decompositions of Biomedical Signals for their Recognition

摘要

The estimation issues of features' informativity in portraits of biomedical signals during their mutually orthogonal decompositions in recognition systems with supervised learning and the technology of revealing and the implementation of reserves of improving the probability of correct pattern recognition by reducing the uninformative features from their portraits are being considered. The content and impact of the proposed technology for the given signals' decomposition are being illustrated by the test case in the application to the task of QRS-complexes types' recognition, suitable in the current study by the availability of existing data for recognition system training.