Algorithms for Collective Decision-Making

Course Information

What this course is about

This course will explore the mathematical foundations of collective decision-making problems, i.e., when a group decision is made based on the preferences of individual agents. These problems arise in a variety of settings, including day-to-day tasks such as a group of friends deciding how to fairly split the rent for their shared apartment as well as high-stakes scenarios such as political elections and matching applicants with jobs or organ donors with patients.

We will use tools and techniques from theoretical computer science and artificial intelligence to design and analyze algorithms for collective decision-making problems. The narrative will center around economic issues such as fairness, efficiency, and incentives and how they interact with computation.

Who is this course aimed at

Students interested in research on algorithmic aspects of economics and game theory should find the course relevant.

Prerequisites

This is a theoretical course that will require mathematical maturity (in particular, the ability to understand and write formal mathematical proofs), and a strong grasp of algorithms and computational complexity (so, COL351 or an equivalent course). We will not assume any background in economics.

Reference text

Frequent references will be Handbook of Computational Social Choice (available in IITD library) and Trends in Computational Social Choice, although individual lectures will refer to different papers that will be shared in the table below.

Admin

Lecture timings+venue

Tue,Fri 3:30-5PM at IV LT-1 (Block IV). Location on Google Maps.
Announcements will be made on the Microsoft Teams channel.

Office hours

By email.

Evaluation policy

The evaluation will be based on four assignments (10% each), class participation (20%), and a research project (40%). Depending on the class size, it may be possible to do the assignments and project in groups.

For the project, you will be expected to read a paper (or a set of closely related papers); a suggestive list is here. The deliverables will include a presentation and a report. There will also be a mid-term project update meeting.

Your final presentation and report should clearly convey the problem studied by the paper(s), the status of prior work, the main contribution of the paper(s) and why you think it is novel and interesting, as well as directions for future research. There will be a significant weightage for new observations and insights, so you are strongly encouraged to present those as well.

How to register

Send an email to Rohit with information about your prior coursework and relevant research experience.