AI Department Scientists Presented Research Results 

30 August 2023

On 29 August, Springer published the book “System Analysis and Artificial Intelligence” in the series Studies in Computational Intelligence (SCI, volume 1107), edited by M. Zgurovsky and N. Pankratova.

This book contains the latest scientific work of Ukrainian scientists and their colleagues from other countries of the world in three interrelated areas: systems analysis, artificial intelligence and data mining. The included articles present the theoretical foundations and practical applications of the latest tools and methods of artificial intelligence, scenario planning, decision making and computational intelligence for important areas of human activity.

The results of research at Artificial Intelligence Department are presented in chapter:

Bodyanskiy, Y., Chala, O., Filatov, V., Pliss, I. (2023). Neo-Fuzzy Radial-Basis Function Neural Network and Its Combined Learning. In: Zgurovsky, M., Pankratova, N. (eds) System Analysis and Artificial Intelligence . Studies in Computational Intelligence, vol 1107. Springer, Cham. https://doi.org/10.1007/978-3-031-37450-0_19

The Neo-Fuzzy Radial Basis Function Stacking Neural Network (NF-RBFNN) is a proposed hybrid neural network system that combines traditional RBFNN and neo-fuzzy neuron. The combination aims to create an efficient system that performs well in scenarios with limited data sets or when fast learning is required due to online data acquisition. NF-RBFNN consists of two independent subsystems, which facilitate quick parameter setting and implementation. With the ability to approximate nonlinear functions and handle uncertain information, the neo-fuzzy neuron and traditional radial basis function neural network are known for their respective strengths. By merging these two approaches, the NF-RBFNN is an effective hybrid neural network that can improve approximation properties while handling uncertainties. Ultimately, this simple yet efficient system excels in situations where quick learning and uncertain information handling come into play. With its impressive performance and straightforward implementation, the system holds great potential for a multitude of practical applications in various fields.

We sincerely congratulate the authors on publishing in Springer!