Research

I am passionate about fundamental principles that underlie randomness of data and algorithms found in networks and applying those in various domains such as smart energy systems, distributed and multimodal learning. Some specific topics that I work on include:

• large deviations principle for distributed inference

• high probability and large deviations bounds for machine learning

• large deviations for consensus and products of stochastic matrices 

• federated and one-shot learning

• multimodal learning

• P2P energy trading

• sensor selection

PhD students:

• Nemanja Petrović

• Vladimir Jankov

MSc students:

• Nemanja Petrović, "Consensus of linear dynamical systems over Markov chain matrices on directed topologies", Data science MSc program, Department of mathematics and informatics, Faculty of sciences, University of Novi Sad

• Vladimir Jankov, "DDoS attack detection and classification with stochastic block models", Department of energetics, electronics and communications, Faculty of technical sciences, University of Novi Sad 

• Kristina Licenberger, "Semidefinite Programming Approach to Visual Storyline Creation", Data science MSc program, Department of mathematics and informatics, Faculty of sciences, University of Novi Sad; finalist for the best MSc thesis at the Department of informatics, Faculty of Sciences, University in Novi Sad

• Jelena Mihajlović, "Detecting communities in networks using belief propagation algorithm", Data science MSc program, Department of mathematics and informatics, Faculty of sciences, University of Novi Sad

• Jelena Novaković, "Data mining with privacy guarantees: Decision trees with differential privacy", Data science MSc program, Department of mathematics and informatics, Faculty of sciences, University of Novi Sad