MISANU

PROGRAM

Plan rada Seminara iz veštačke inteligencije za MAJ 2023.

Registraciona forma za učesće, i link na predavanje ako ste već registrovani:
https://miteam.mi.sanu.ac.rs/asset/CW5nJWDSEZDj7p32p
Ukoliko želite samo da gledate predavanje bez mogućnosti aktivnog učešća, prenos će biti dostupan na:
https://miteam.mi.sanu.ac.rs/asset/4LNW8WtML7rLKojoz
Na ovom linku se mogu pronaci kratka uputstva na srpskom i engleskom:
https://miteam.mi.sanu.ac.rs/asset/Kc7qJtEvoMFx9MFnz

SREDA, 03.05.2023. u 19:00, Online
Vladimir Kostić, Numerical Linear Algebra at Faculty of Science, University of Novi Sad; Computational Statistics and Machine Learning group at Istituto Italiano di Tecnologia, Genova, Italy
AI FOR SCIENCE: STATISTICAL LEARNING PERSPECTIVE TO KOOPMAN OPERATOR THEORY FOR DATA-DRIVEN DYNAMICAL SYSTEMS
We are witnessing striking progress originating from the use of AI, and, in particular, Machine Learning (ML) technologies in science. Traditional scientific modelling by equations of motion is being more and more powered by complementary data-driven approaches to solve challenging problems such as protein folding and fluid-dynamics. In this quickly growing field, the theory of Koopman operators have found particularly prominent place due to the fact that non-linear dynamical systems can be handily described by the associated linear (Koopman) operators whose action evolves observables of a system forward in time. While data-driven algorithms to reconstruct such operators are now well known, their relationship with statistical learning is still largely unexplored. To bridge this gap, in this talk we will present a framework to learn Koopman operators from finite data trajectories using reproducing kernel Hilbert spaces (RKHS). Doing this, we provide high-probability finite-sample theoretical learning guarantees, which are of the paramount importance for safe and trustworthy employment of the AI based on the Koopman operator theory in scientific applications.

SREDA, 10.05.2023. u 19:00, Online
Zoran Obradovic, Data Analytics and Biomedical Informatics Center, Computer and Information Sciences Department, Statistics Department, Temple University, Philadelphia, USA
PREDICTIVE ANALYTICS FOR CLINICAL DECISION MAKING
An overview of our machine learning research aimed to facilitate decision making in healthcare will be presented in this talk. Challenges will be discussed related to integration of biomedical knowledge and heterogeneous medical records, learning from censored observations, knowledge discovery in large temporal data as well as in data obtained across multiple smaller studies. Examples of our proposed solution will be shown in the context of disease diagnosis and progression prediction in Alzheimer's disease, Anxiety disorder, Cancer, Chronic Kidney Disease and Diabetes.

SREDA, 17.05.2023. u 20:15, Online
Stevan Gostojić, Fakultet tehničkih nauka, Univerzitet u Novom Sadu
PRIHVATLJIVOST VEŠTAČKE INTELIGENCIJE U DIGITALNOJ FORENZICI
Veštačka inteligencija je postala ključna tehnologija, koja pronalazi sve više primena u društvu. Oblast u kojoj VI može da pokaže veliki potencijal je digitalna forenzika (tj. stručna disciplina čiji predmet su identifikacija, prikupljanje, čuvanje, pregledanje, analiza i prezentacija digitalnih dokaza korišćenjem naučno i pravno valjanih metoda i alata). Međutim, postoji više izazova u vezi sa primenom tehnika i alata VI u digitalnoj forenzici. Ova studija ima za cilj da analizira upotrebu tehnika i alata VI u oblasti digitalne forenzike i da pruži preporuke koje mogu da utiču na razvoj tehnika i alata VI i na donošenje propisa koji regulišu njihovu upotrebu, da bi se osigurala etička i efektivna upotreba VI.

SREDA, 24.05.2022. u 19:00, Online
Dušan Ramljak, Penn State Great Valley, Pennsylvania State University, USA
POTENCIJAL VESTACKE INTELIGENCIJE U BORBI PROTIV ALKOHOLIZMA I NARKOMANIJE
Trenutno se pretpostavlja da ima oko 40 miliona ljudi u SAD koji imaju probleme prouzrokovane prekomernim korišćenjem alkohola i droga. Nerazdvojni činilac borbe za prevenciju i tretman tih i takvih problema je upotreba širokog spektra postojećih podataka o zdravstvenom, fizičkom i psihološkom stanju ljudi kod kojih se problemi manifestuju. Naše istraživanje pokušava da postavi osnove metoda i otkrije mogućnosti veštačke inteligencije da pomogne u tretmanu pomenutih problema. Važan deo istraživanja je evaluacija efikasnosti metoda koje se baziraju na razumevanju uzročno posledičnih veza iz podataka koje generiše online platforma za borbu protiv alkoholizma i narkomanije. Platforma Smart Personalized Adaptive Recovery System SPARx je prva potpuno virtuelna platforma za tretman alkoholizma i narkomanije koja pruža usklađivanje tretmana na osnovu razumevanja navika pacijenata koje tretira. Predavanje će biti održano na srpskom jeziku koliko god bude moguće s obzirom da se bazira na radu studenata koji koriste podatke iz SAD firme koja je napravila platformu upotpunjene podacima američke nacionalne ankete o korišćenju droga i zdravlju korisnika alkohola i opojnih droga.

SREDA, 31.05.2022. u 19:00, Online
José Pedro Cabalar Fernández, Computer Science and Information Technology Department, University of A Coruña, Spain
EXPLANATIONS FOR PRACTICAL KNOWLEDGE REPRESENTATION
Achieving a truly explainable Artificial Intelligence (AI) is nowadays a challenge not only for machine learning "black box" techniques but even for symbolic automated reasoning, especially when the explanations of a conclusion are too large or hard to follow for an average human user. In the talk, I will first discuss several aspects of explainable symbolic AI systems and then proceed to introduce a kind of explanations for logic programs based on so-called justified models, that is, models for which an explanation graph exists. Finally, I will illustrate these concepts with several examples using the tool xclingo, that generates explanations for logic programs under the Answer Set Programming paradigm.

Ovaj onlajn seminar nastao je kao nastavak sastanka “Serbian AI Meeting” i zamišljen je da na njemu istraživači iz Srbije i iz dijaspore, kao i istraživači sa univerzteta, naučnih instituta i iz prakse predstavljaju naučne teme i rezultate iz oblasti veštačke inteligencije.
Link za svako pojedinačno predavanje biće dostavljen dan pre održavanja predavanja.

Andreja Tepavčević
Rukovodilac seminara