Registration Number
0220U103442,
0117U003920
, R & D reports
Title
Methods and tools of decision-making for data processing in intellectual pattern recognition system
popup.stage_title
Head
Lovkin Vаlerii M., Кандидат технічних наук
Registration Date
06-10-2020
Organization
Zaporizhzhya National Technical University
popup.description2
The object of the research is pattern recognition. The research objective is to develop and research methods and tools of decision making for data processing in intelligence systems of pattern recognition. The methods of research include methods of artificial intelligence, pattern recognition, decision theory methods, highly productive computations. Methods of data sample reduction, which allow to reduce training data sets, eliminating redundant and uninformative features and observations of samples, have been developed. Methods for building of models based on deep learning neural networks have been proposed. The method for the synthesis of neuromodels based on multimodal learning and discriminant tuning has been developed. The method for constructing cognitive neural networks based on multimodal learning with the simultaneous usage of several models with different structure has been developed. The parallel method for the synthesis of recognition models have been proposed. It uses probabilistic optimization to adjust the parameters of the synthesized models. The initial set of solutions is formed taking into account the information about the training sample. The method for relearning of synthesized recognition models has been developed. Information technology model and software have been developed to solve image recognition problems. The results of the research work are recommended for use when problems of pattern recognition, decision making, etc. are solved.
Product Description
popup.authors
Il'yashenko Matvij B.
Gofman Yevgeniy O
Liovkin Valerii M
Oliinyk Andrii O.
Olenchenko Ihor I
popup.nrat_date
2021-03-17
