Naredni sastanak Seminara biće održan onlajn u utorak, 21. oktobra 2025, u sali 301f Matematičkog instituta SANU sa početkom u 14:15.
Predavač: Tian Lan, Electrical and Computer Engineering Department, School of Engineering and Applied Science, George Eashington University, USA
Naslov predavanja: MULTI-AGENT DECISION-MAKING IN THE OPEN WORLD
Apstrakt:
AI/ML has demonstrated tremendous success in many challenging tasks with superhuman performance. Many decision-making problems in the open world are too complex to be modeled as a monolithic single-agent problem and
thus require the formulation of multi-agent decision making. In this talk, we present some recent research progress on multi-agent decision-making in the open world. We will discuss some key challenges in this space, such as
collaborative decision making, multi-agent skill discovery, game-theoretical formulations, and human-AI collaboration. We will also demonstrate several applications of the results to real-world network infrastructures, such as datacenter resource reservation and 5G network management.
Biografija predavača: Tian Lan is a professor at the Department of ECE, George Washington University in DC and the Director of Human-centric Autonomy and Robotics (HART) lab. He received PhD in Electrical Engineering from the Princeton University. His research interests include machine learning, optimization, and relevant applications to networking and cyber security. The research is currently being supported by NSF, DARPA, ONR, ARO, USMA, Meta, and CISCO. He has received 6 best paper awards (e.g., IEEE Signal Processing Society, INFOCOM, Globecom, and Mobihoc), 6 industry research awards (from AT&T, CISCO, and META), as well as several faculty recognition and innovation awards. He is currently serving as a member of FCC Technological Advisory Council (TAC), INFOCOM 2026 TPC Co-Chair, Fellow of National Quantum Lab at UMD (NQL), and Associate Editor for IEEE/ACM Transactions on Networking.
Napomena: Predavanja možete pratiti na daljinu. Sve informacije su dostupne na stranici:
https://miteam.mi.sanu.ac.rs/asset/qGapAHyEBad2FDwXR
