Aktualny program
Aktualny program
Piątki, godz. 11:00, sala 44 (Laboratorium Mi^2)
- 30 października 2026, Lucian Coroianu (Dep. of Mathematics and Informatics, University of Oradea, Romania), Approximations by neural network operators activated by sigmoidal functions
In 2013 Costarelli and Spigler introduced Neural Network (NN) linear and positive approximation operators activated by sigmoidal functions. Recently, we introduced NN linear and positive operators of Durrmeyer type, generalizing the discrete and the Kantorovich variants. In this talk I will present results on the uniform convergence and L^p norm convergence of these operators, including quantitative estimates using the Hardy-Littlewood maximal function. In addition, our studies on NN operators inspired us to obtain estimations for the absolute central moments of some approximation operators. Some of these new results will also be part of the presentation. - 16 października 2026, Shivani Singh (Dep. of Computing Science, Umeå University, Sweden), Feature selection for set-valued data using fuzzy rough sets
We propose a fuzzy rough set-based feature selection approach using a tolerance relation for set-valued data, where objects may be described by multiple possible attribute values. The method exploits fuzzy rough approximations to evaluate feature relevance while preserving uncertainty and imprecision inherent in set-valued information, enabling the selection of a compact and informative feature subset.