Nonna Kulishova

Dr. Nonna Kulishova

orcid

Research Focus & Metrics

Academic Career Length

19Years (2008-Present)

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

Representative Publications

View All Publications
8 September 2017 in International Journal of Intelligent Systems and Applications
Verified
Journal Article
Open access
PDF file
ZHZhengbing Hu
Zhengbing Hu
YBYevgeniy V. Bodyanskiy
Yevgeniy V. Bodyanskiy
NKNonna Kulishova
Nonna Kulishova
OTOleksii K. Tyshchenko
Oleksii K. Tyshchenko
An article introduces a modified architecture of the neo-fuzzy neuron, also known as a "multidimensional extended neo-fuzzy neuron" (MENFN), for the face recognition problems. This architecture is marked by enhanced approximating capabilities. A characteristic property of the MENFN is also its computational plainness in comparison with neuro-fuzzy systems and neural networks. These qualities of the proposed system make it effectual for solving the image recognition problems. An introduced MENFN's adaptive learning algorithm allows solving classification problems in a real-time fashion.

9 December 2018 in Data
Verified
Journal Article
Open access
PDF file
YBYevgeniy Bodyanskiy
Yevgeniy Bodyanskiy
NKNonna Kulishova
Nonna Kulishova
OCOlha Chala
Olha Chala
Methods of machine learning and data mining are becoming the cornerstone in information technologies with real-time image and video recognition methods getting more and more attention. While computational system architectures are getting larger and more complex, their learning methods call for changes, as training datasets often reach tens and hundreds of thousands of samples, therefore increasing the learning time of such systems. It is possible to reduce computational costs by tuning the system structure to allow fast, high accuracy learning algorithms to be applied. This paper proposes a system based on extended multidimensional neo-fuzzy units and its learning algorithm designed for data streams processing tasks. The proposed learning algorithm, based on the information entropy criterion, has significantly improved the system approximating capabilities. Experiments have confirmed the efficiency of the proposed system in solving real-time video stream recognition tasks.

6 October 2016
Verified
Conference Paper
Closed access
NKNonna Kulishova
Nonna Kulishova
The problem of emotion state recognition using the sigma-pi artificial neural network is considered. The specific feature of the considered network is the presence of two different types of activation functions: sigmoid and bell-shaped. A learning algorithm for the sigma-pi network is proposed. This algorithm is characterized by high approximation accuracy especially for nonlinear processes in real time. Training and testing network on image dataset was made.

1 August 2018
Verified
Conference Paper
Closed access
YBYevgeniy Bodyanskiy
Yevgeniy Bodyanskiy
NKNonna Kulishova
Nonna Kulishova
DMDaria Malysheva
Daria Malysheva
Many tasks require human facial expressions automatic recognition in real time. Recent solutions to this problem using machine learning methods have been based on the applying of training data sets that include hundreds of thousands of samples. The formation of these data is too costly. In this paper, the architecture of a system using extended neo-fuzzy neurons for online emotions recognition is examined. We propose the algorithm which is based on the entropy criterion for learning the system and reducing the amount of training data thousands of times.

7 February 2025 in Journal of Imaging
Verified
Journal Article
Open access
PDF file
NKNonna Kulishova
Nonna Kulishova
DAIVA SAJEKDAIVA SAJEK
DAIVA SAJEK
The rapid development of machine learning and artificial intelligence approaches is finding ever wider application in various areas of life. This paper considers the problem of improving editorial and publishing processes, namely self-publishing, when designing book covers using machine learning and generative artificial intelligence (GAI) methods. When choosing a book, readers often have certain expectations regarding the design of the publication, including the color of the cover. These expectations can be called color preferences, and they can depend on the genre of the book, its target audience, and even personal associations. Cultural context can also influence color choice, as certain colors can symbolize different emotions or moods in different cultures. Cluster analysis of book cover images of the same genre allows us to identify color preferences inherent in the genre, which is proposed to be used when designing new covers. The capabilities of generative services for creating and improving cover designs are also investigated. An improved flow chart for using GAI in creating book covers in the process of self-publishing is proposed, which includes new stages, namely exploring, conditioning, and evolving. At these stages, the designer creates prompts for GAI and examines how they and GAI’s issuances correspond to the task. Conditioning allows for even more precise adjustment of prompts to features of each book, and the evolving stage also includes post-processing of results already received from GAI. Post-processing, in turn, can be performed both in generative services and by a designer. The experiment allowed us to use the machine-learning method to determine which colors are most often found in book cover layouts of one of the genres and to check whether these colors correspond to harmonious color palettes. In accordance with the proposed scheme of the design process using generative artificial intelligence, versions of book cover layouts of a given genre were obtained.

1 September 2015
Verified
Conference Paper
Closed access
NKNonna Kulishova
Nonna Kulishova
YBYevgeniy Bodyanskiy
Yevgeniy Bodyanskiy
The paper presents the architecture of the color digital image filter, which is based on fuzzy peer group and uses a flexible two-dimensional membership functions. This allows to make filtering more variability in the conditions of action of distortion different types on the images with a different structure.