Course information
- Term: Fall 2026
- Lecture time: Tuesday/Thursday, 9–10:15am
- Lecture location: Ames 234
- Instructor: Nico Christianson
- Office hours: Thursday 10:20–11:20am or by appointment, Malone 223
Description
Artificial intelligence and machine learning hold significant promise for improving algorithmic decision-making across domains. This course will survey recent advances in integrating AI/ML models into algorithm design. We will focus on two different paradigms: algorithms with predictions, which seek to leverage black-box, potentially unreliable predictions to improve performance while maintaining robustness; and learning-based approaches, where AI/ML models are trained to directly perform algorithmic reasoning. Throughout, we will emphasize settings where provable guarantees—such as robustness to prediction error and generalization bounds—can be obtained.
Topics will include:
- online and approximation algorithms with machine-learned predictions;
- data-driven algorithm design and selection; and
- machine learning for algorithmic reasoning.
Prerequisites
Mathematical maturity (i.e., familiarity with proofs) and prior courses in algorithms (EN.601.433/633 or equivalent) and machine learning (EN.601.475/675 or equivalent) will be expected.
Course activities
- Lectures: Lectures will cover recent papers related to the course topic, and will be focused on proving the main technical results in each paper.
- Readings and online discussion posts: Students should read each week’s paper(s) before that week’s lectures to become familiar with the problem setting, motivation, and high-level main results. There is no need to read and understand the proofs in depth, as this will be the focus of lecture time. A reading-response prompt will be due on the Canvas discussion board each Monday. Students should write a 2-3 sentence response describing, e.g., their thoughts on the paper, its strengths and weaknesses, any questions about the results, or ideas for follow-up work.
- Problem set: One problem set will be assigned in the first two weeks of class, covering foundational topics that will be useful throughout the course.
- Paper presentations: During the last few weeks of class, students will present (in groups of 2-3) recent research papers related to the course topic.
- Project: Students will complete (in groups of 1-3) a semester-long research project related to the course topics and their interests.
Important dates
- Monday evenings: Reading responses due
- September 17: Problem set due
- October 8: Project proposal due
- November 5: Mid-semester project update presentations
- November 17 – December 10: Paper presentations
- December 15, 9am-12pm: Final project presentations (writeup due by end of day)
Additional resources
- https://home.ttic.edu/~avrim/MLforAD25/index.html, a related course taught at TTIC in 2025
- https://vitercik.github.io/ml4algs/ and https://vitercik.github.io/ai4algs_25/, related courses taught at Stanford