Activities
Where we are today?
The project has enabled us to build a truly interdisciplinary team, which combines diverse expertise in a way that keeps focusing on the observed problem holistically while advancing each of the involved scientific fields. While we expect the resulting scientific output to have a significant impact, we also find the interdisciplinary research environment developed during the project of very high importance.
Zoom Meetings
As it was not possible to travel to Sweden in the period 31.12.2020. - 31.12.2021, the collaboration was made active, through holding regular and fruitful ZOOM meetings. Project participants discussed the topics and directions in which the research will take place.
A research team from Serbia visited Uppsala!
The first visit in Jun 2022 - Short overview!
The scientific goals of the project are related to the application of algebra and logic in machine learning and artificial intelligence. In the period that preceded the travel, participants analyzed the state-of-the-art of this rapidly developing field of research. The main areas that attracted the attention of participants are graph neural networks (GNNs), the application of GNNs to medical image processing, and explainable artificial intelligence.
The meeting gathered participants in order to discuss and more precisely define open problems and further plans. Project participants from Novi Sad have shown interest in the theoretical investigation of spectral graph theory and its application to graph neural networks.
Partners from Uppsala already had notable experience in applying convolutional neural networks (CNNs) in image analysis.
Since GNNs are a generalisation of CNNs, partners from Uppsala expressed excitement about the possibility of applying GNNs to the analysis and classification of medical images.
In the further period, all participants will investigate existing GNNs.
Participants from Novi Sad will investigate how other (previously not considered in GNNs) spectral transformations can influence GNNs and some specific applications.
The second visit in November 2022- Short overview!
During the visit, participants put considerable effort into seeking topics of joint interest that would motivate further collaboration.
The final intersection of topics of interest converged to those related to the development and application of graphs-based methods, and in particular graph neural networks, in biomedical image data analysis.
In particular, we have focused on the identification of characteristics that can differentiate graphs that describe the distribution of immune cells in the tumor microenvironment which corresponds to successful vs unsuccessful cancer therapy.
During the meeting at Uppsala University, Ivan Prokić gave a scientific lecture, with the title “Theoretical Foundations for Correct Formation of Distributed Clouds”;
During the meeting at Uppsala University, Jovana Dedeić gave a scientific lecture, with the title “Formal Techniques for Modeling and Analysis of Software Systems for Transactions and Dynamic Update”;
The expertise of participants from University of Novi Sad will stay along the lines of investigating theoretical aspects of the problem, which is in the first place focused on the study of exploration and identification of suitable known properties of graphs that are of crucial importance to this particular application area.