CP 270: Multi-Agent Systems, Control and Learning (3:0)

Announcements

First class: Tuesday, 4 August

Time and venue

Class timings: Tuesdays, Thursdays 15:30-17:00
Venue: TCS Smart X Hub G12

Instructor

Pavan Tallapragada
Office: TCS Smart X Hub 422

email: pavant@iisc.ac.in (prefix “MAS”, without quotes, in email subject)

Course overview

Broadly, the course contains a collection of core topics and breadth topics. The core topics introduce fundamental frameworks and most of them would be covered at a good level of depth. The breadth topics are aimed at widening the breadth of students’ awareness and would be covered selectively and at a shallow level.

Course objectives/expected outcomes:

Cover some fundamental topics in the area. Provide an exposure to the breadth of topics in the area. Provide depth on selected topics through course projects and discussions. Help students learn to read and verify scientific literature; and learn technical presentation skills.

Course outline

Core topics:

Intro to graph theory
Consensus dynamics and its various applications
Social networks, opinion dynamics, social choice theory
Distributed optimization, federated learning
Intro to game theory, network games, population games, population dynamics, game dynamics and learning in games, mechanism design
Intro to multi agent reinforcement learning
Graph neural networks
Applications in engineering systems such as in robotics, transportation systems, power systems

Breadth topics: Synchronization, formation control, distributed estimation, coverage control, connectivity maintenance, blockchain; privacy and security in networked and distributed control systems; other distributed computing and control applications; network formation; control over communication networks, event-triggered control; brain networks, ecological systems; compartmental models, epidemic spread; mixture of experts in machine learning, agentic AI, LLM councils; social networks of humans and AI agents.

Projects and presentations: A major project and a couple of minor projects and presentations would enable the students to gain a deeper understanding of specific areas of interest. A couple of the projects would also be organized as hackathons or group competitions.

Prerequisites

Aptitude and interest in mathematical modeling and analysis. While there are no particular prerequisite courses, students should have taken or should currently be taking some “mathematical” courses at IISc such as on linear algebra, probability, random processes, optimization, linear or nonlinear dynamical systems. Instructor would decide on each student’s enrollment on a case by case basis.

Grading

Grading would be based on homeworks, minor projects, one mid-term exam, literature reading, presentations and a major project.

Resources

Most of the core topics would be taught from reference books, review papers and current research literature. Breadth topics would primarily be based on review papers and current research literature.

Bullo, Francesco. Lectures on Network Systems
Mesbahi, Mehran, and Magnus Egerstedt. Graph theoretic methods in multiagent networks. Princeton University Press, 2010.
Bollobás, Béla. Modern graph theory. Vol. 184. Springer Science & Business Media, 1998.
Barabasi, Albert-Laszlo, Network Science. Cambridge University Press, 2016.
Shoham, Yoav, and Kevin Leyton-Brown. Multiagent systems: Algorithmic, game-theoretic, and logical foundations. Cambridge University Press, 2008.
Narahari, Yadati. Game theory and mechanism design. Vol. 4. World Scientific, 2014.
Sandholm, William H. Population games and evolutionary dynamics. MIT press, 2010.
Fudenberg, Drew, and David K. Levine. The theory of learning in games. Vol. 2. MIT press, 1998.
Albrecht, Stefano V., Filippos Christianos, and Lukas Schäfer. Multi-agent reinforcement learning: Foundations and modern approaches. MIT Press, 2024.
Current literature

Academic integrity

Please be aware of the IISc academic integrity policy. Any violation of it will be dealt with strictly.