Predavanja možete pratiti i online putem MITEAM stranice Seminara za računarstvo i primenjenu matematiku:
https://miteam.mi.sanu.ac.rs/asset/qGapAHyEBad2FDwXR
PLAN RADA SEMINARA ZA NOVEMBAR 2025. GODINE
Utorak, 04.11.2025. u 14:15, Knez Mihailova 36, sala 301f i
Online
Iva Vojinović, Fakultet organizacionih nauka, Univerzitet u Beogradu
SERVISI PAMETNOG GRADA ZA INKLUZIJU OSOBA SA INVALIDITETOM I STARIH LICA
Problem koji se razmatra u ovom predavanju su servisi pametnih okruženja koji na održivi način podržavaju inkluziju osoba sa invaliditetom, pružajući im pogodnosti i olakšavajući svakodnevni život. Razmatrane su kombinacije pametnih servisa i postojećih inteligentnih rešenja koje pružaju najveću vrednost osobama sa invaliditetom za samostalno obavljanje svakodnevnih aktivnosti u urbanoj sredini. Ispitana je spremnost za korišćenje održivih inkluzivnih pametnih servisa od strane osoba sa invaliditetom i starih lica. Rezultati ukazuju na pozitivnu sklonost ka uslugama pametnih gradova u svim demografskim grupama. Faktori jednostavnosti korišćenja i mogućnosti za društveno angažovanje identifikovani su kao ključni pokretači usvajanja pametnih tehnologija u razmatranom kontekstu.
Utorak, 18.11.2025. u 14:15, Knez Mihailova 36, sala 301f i
Online
Daniel Rodríguez García, Computer Science Department of the University of Alcala, Madrid, Spain
THE INFLUENCE OF DATA COMPLEXITY ON ALGORITHMIC RELIABILITY IN SOFTWARE DEFECT PREDICTION
We explore the role of data complexity in software defect prediction, focusing on how structural characteristics of datasets influence both machine learning performance and feature selection reliability. Using a set of complexity metrics and multiple algorithms, the study reveals that certain data traits, such as dimensionality, overlap, and network structure, can significantly affect predictive stability and effectiveness. Knowing the complexity of the data can offer practical insights for selecting appropriate models and feature selection strategies based on dataset properties, contributing to more robust and interpretable defect prediction systems.
Utorak, 25.11.2025. u 14:15, Online
Robert Doža, Matematički fakultet, Beograd
OPTIMIZATION APPROACH TO THE PROBLEM OF SPECTRAL RECONSTRUCTION OF GRAPHS
Optimization problems on graphs consists of finding graphs (or a family of graphs) that minimize or maximize a given invariant (or some combination of the considered invariants). These problems usually belong to the NP-hard class making the exhaustive enumeration of all possibilities a unacceptable approach. Efficient optimization algorithms developed for these problems can help theoreticians to make hypothesis and get some useful hints how to analytically prove them. We consider a spectral reconstruction of graphs (SRG), an open problem from the spectral graph theory. It is a known fact that graphs are not uniquely determined (characterized) by their spectra in the general case. This means that, for certain graphs, there could exist several cospectral mates, i.e., non-isomorphic graphs corresponding to the same spectrum. The objective of SRG is to find at least one graph with a spectrum equal to a given constant vector. This problem is very important and has numerous real-life applications. A common approach to perform incomplete search in SRG is to develop metaheuristic tailored for this particular problem. Metaheuristic methods are general-purpose, stochastic, incomplete search algorithms to address NP-hard optimization problems. We develop several approaches to SRG, based on Variable Neighborhood Search (VNS) metaheuristic, taking into account the graph properties defined by its spectrum. We evaluate the performance of the proposed algorithms on the reconstruction of some well-known graphs and their potential non-isomorphic cospectral mates. Besides providing a graph for a given spectrum, the proposed algorithms can find cospectral mates, indicating that the corresponding graphs are not characterized by the spectrum. However, if a single graph is found, we cannot conclude that it is unique for a given spectrum.
RUKOVODIOCI SEMINARA
MI SANU
Vera Kovačević-Vujčić
Milan Dražić
FON
Zorica Bogdanovic
Marijana Despotovic-Zrakic
IEEE
Bozidar Radenkovic
