CMSC 35401: Topics in Machine Learning

The Interplay of Economics and Machine Learning (Winter 2024)

---

Basic Information

Class Location: Ryerson Physical Laboratory 255
Class Time: Thursdays, 2:00pm to 4:50pm, with 15-mins middle break
Instructor: Haifeng Xu

• Email: haifengxu AT uchicago.edu
• Office: Crerar 260
• Office hours: Thursday 4:50 - 5:50 pm (rightly after class)

TA: Alec Sun

• Email: alecsun@uchicago.edu
• Office Hours: Tue 4:30 to 6 pm; Room: JCL 205

Course Material: There will not be any official textbook, but the slides and links to reading materials will be posted on the course schedule after each lecture.

Announcements

• Dec 27: Course website is up!
• Jan 11: First HW is out; due 01/20.
• Jan 20: Second HW is out; due Feb 3rd.
• Jan 20: Projection instruction is out; proposal is due also on Jan 27 (i.e., in a week).



Course Description

Welcome! This is a PhD-level course covering topics at the interface between machine learning and economics. In many economic applications, the problem either is too complex or has much uncertainty. In such cases, machine learning approaches help to design more realistic or practical algorithms. Conversely, in many application of machine learning, the algorithms have to interact with economic agents, such as self-interested data suppliers whose objectives are not aligned with the algorithm or human content creators (e.g., Youtubers) who are now challenged by generative AI technology. These problems form an intriguing interplay between machine learning and economics, and have attracted a lot of recent research attention. This course will discuss several recent research directions in this space. Our goal is to cover (some selected) basic results in the following four directions. Alone the way, we will also cover necessary basics such as game theory, learning theory and information economics.

Topics covered in this course, and tentative syllabus:

• First Half: Machine Learning for Economic Problems
  • (week 1) Linear programming and duality
  • (week 2-3) Intro to game theory, zero-sum games
  • (week 3-4) No regret learning and its convergence to equilibrium
  • (week 5) Bandit algorithms
• Second Half: Economic Aspects for Machine Learning
  • (week 5-6) Information, valuation of information and data
  • (week 7-8) Learning from strategic data sources and distribution shifting
  • (week 9) Economics of generative AIs, human-vs-AI content creation competition
This is primarily a theory course, and lecture-based. That said, we will focus on proofs and will not write code. Prerequisites include linear algebra, probability and algorithms. Familiarity to optimization is a plus but not necessary. Note that no background on game theory or learning theory is required.


Tentative Schedule and Readings

Lec No.	Lectures	Readings
	PART I:	Learning for Economic Problems
1 (Jan 4: I)	Introduction [slides]	Kleinberg/Leighton paper
2 (Jan 4: II)	Basics of LPs [slides]	Chapter 2.1, 2.2, 4.3 of Convex Optimization by Boyd and Vandenberghe
3 (Jan 11: I)	LP duality [slides]	Lecture notes 5 and 6 of an optimization course by Trevisan
4 (Jan 11: II)	Intro to Game Theory (I) [slides]	Section 3.1, 3.2, 3.3 of an game theory book by Shoham and Leyton-Brown
5-6 (Jan 18)	Intro to Game Theory (II) [slides]	Equilibrium analysis of GANs by Arora et al.
7 (Jan 25: I)	Intro to Online Learning [slides]
8 (Jan 25: II)	Multiplicative Weight [slides]	A survey paper on MWU and its applications by Arora et al.
9 (Feb 1: I)	Swap Regret [slides]	A note by Balcan on converting regret to swap regret
10 (Feb 1: II)	Multi-Armed Bandits [slides]	Section 2, 3 of the Book by Bubeck and Cesa-Bianch on Bandits
	PART II:	Economic Aspects of Machine Learning
11 (Feb 8: I)	Information Design [slides]	Bayesian Persuasion and Information Design paper
12 (Feb 8: II)	Pricing of Information [slides]	Quantifying information and Optimal Pricing of Information
13 (Feb 15: I)	Strategic Learning I [slides]	PAC-learning for Strategic Classification paper
14 (Feb 15: II)	Strategic Learning II [slides]	How Do Classifiers Induce Agents To Invest Effort Strategically?
16 (Feb 22: II)	Performative Prediction [slides]	Performative Prediction: Past and Future
15 (Feb 22: I)	Tradeoffs of Fairness [slides]	Inherent Trade-Offs in the Fair Determination
17 (Feb 29: I)	Economics of Generative AI [slides]	Human vs. Generative AI in Content Creation Competition
18 (Feb 29: II)	Project [Schedule]




Homework

Due date	Homework	Note
01/20	Homework 1	Here is a HW solution template in case you need one
02/03	Homework 2	Good luck!
02/17	Homework 3	Good luck!
03/02	Homework 4	Good luck!




Requirements and Grading

Participation counts for 10%. Homework assignments will count for 45% of the grade. There will be 4 proof-based assignments, roughly 2 weeks apart each. While discussion among students is allowed, students must write up the solutions independently.
The remaining 45% of the grade is allocated to a research project. Students can team up for the project and each team should consist of 2 to 4 people. The project needs to raise novel technical questions related to the course topics and provide some nontrivial answers.
There is no midterm or final for the course.
For those who take the course as elective: this course can be taken as an elective. To do so, you will need to complete the 1-2 bonus (often more challenging) questions in each assignment.
Late Homework Policy : Each student is allowed one late homework for at most two days from the due date. You may choose whichever homework to use this chance (or not use it). No additional late homework will be accepted.




Students with disabilities or learning needs

We thrive to create a learning experience that is as accessible as possible. If you anticipate any issues related to the format, materials, or requirements of this course, please meet with me outside of class so we can explore potential options. Students with disabilities may also wish to work with the UChicago Student Disability Services to discuss a range of options to removing barriers in this course, including official accommodations. Please visit their website for information on this process and to apply for services online: disabilities.uchicago.edu. If you have already been approved for accommodations through SDS, please send me your accommodation letter, and meet with me if needed, so we can develop an implementation plan together.