Anastasiia Deineko

Dr. Anastasiia Deineko


Research Focus & Metrics

Academic Career Length

11Years (2016-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 January 2017 in Automatic Control and Computer Sciences
Verified
Journal Article
Closed access
PDF file
YBYe. V. Bodyanskiy
Ye. V. Bodyanskiy
ADAnastasiia Deineko
Anastasiia Deineko
YKYa. V. Kutsenko
Ya. V. Kutsenko
The clustering system based on the evolving general regression neural network and self-organizing map of T.Kohonen, is proposed in the paper. The tuning of system is based on “lazy” learning and self-learning using the principle “Winner takes more” at the same time as neighborhood function the output signal of the hybrid network is used. The system’ implementation is characterized by numerical simplicity. The evolving neural network processes data in an online mode and doesn’t suffer from the curse of dimensionality.

1 May 2018 in Automatic Control and Computer Sciences
Verified
Journal Article
Closed access
PDF file
YBYe. V. Bodyanskiy
Ye. V. Bodyanskiy
ADAnastasiia Deineko
Anastasiia Deineko
FEF. M. Eze
F. M. Eze
The architecture of multilayer kernel clustering neuro-fuzzy system and algorithm of its self-learning are intended for the recovery of overlapped clusters in situations when the streams of...

15 December 2020 in Information Technology and Management Science
Verified
Journal Article
Open access
PDF file
YBYevgeniy Bodyanskiy
Yevgeniy Bodyanskiy
ADAnastasiia Deineko
Anastasiia Deineko
IPIrina Pliss
Irina Pliss
OCOlha Chala
Olha Chala
The probabilistic neuro-fuzzy system to solve the image classification-recognition task is proposed. The considered system is a “hybrid” of Specht’s probabilistic neural network and the neuro-fuzzy system of Takagi-Sugeno-Kang. It is designed to solve tasks in case of overlapping classes. Also, it is supposed that the initial data that are fed on the input of the system can be represented in numerical, rank, and nominal (binary) scales. The tuning of the network is implemented with the modified procedure of lazy learning based on the concept “neurons at data points”. Such a learning approach allows substantially reducing the consumption of time and does not require large amounts of training dataset. The proposed system is easy in computational implementation and characterised by a high classification speed, as well as allows processing information both in batch and online mode.

26 December 2018 in Bionics of Intelligence
Verified
Journal Article
Closed access
YBYe. Bodyankiy
Ye. Bodyankiy
KPK. Petrov
K. Petrov
ADAnastasiia Deineko
Anastasiia Deineko
In the paper proposes an approach to the formation of the architecture and training of the evolving radial-basis neural network (RBFN), which processes the data in sequential online mode. It consists of two stages. The first one is based on the use of the T. Kohonen self-organizing map (SOM), which is responsible for managing the number of RBFN neurons and setting the location of the centers of kernel functions in self-learning mode. The second stage is responsible for its training using the method of extreme learning (ELM). This approach allows to avoid the "curse of dimensionality" in the formation of RBFN and significantly increase the speed of its learning.

20 October 2016 in arXiv
Verified
Preprint
Open access
PDF file
YBYevgeniy V. Bodyanskiy
Yevgeniy V. Bodyanskiy
OTOleksii K. Tyshchenko
Oleksii K. Tyshchenko
ADAnastasiia Deineko
Anastasiia Deineko
An architecture of a new neuro-fuzzy system is proposed. The basic idea of this approach is to tune both synaptic weights and membership functions with the help of the supervised learning and self-learning paradigms. The approach to solving the problem has to do with evolving online neuro-fuzzy systems that can process data under uncertainty conditions. The results prove the effectiveness of the developed architecture and the learning procedure.

15 September 2021
Verified
Conference Paper
Closed access
AAAlina Albasova
Alina Albasova
YBYevgeniy Bodyanskiy
Yevgeniy Bodyanskiy
ADAnastasiia Deineko
Anastasiia Deineko
IPIryna Pliss
Iryna Pliss
A matrix neuron and algorithm for its weights tuning are conceived. This neuron processes matrix input directly without its transformation to vector form. Neural networks based on it allow decreasing the number of tunable weights that increases the speed of information processing.