Seminar on Algorithmic Economics and Machine Learning
This is a new 1-credit course to accompany the long-standing algorithmic economics reading group. We read papers and discuss research topics in algorithmic economics, social choice theory, economic theory, algorithmic game theory, and theoretical machine learning.
See the algorithmic economics page for the time and place! The reading group is open to the public, so please feel free to stop by anytime, whether officially enrolled or not.
Syllabus
Prerequisites: None. Anyone is welcome to join us. As we typically read recent research papers, there is rarely anyone in the room that fully understands. That said, familiarity with any subset of {economics, algorithms, machine learning, optimization, statistics} would help, as well as "mathematical maturity", meaning a grasp of proof writing and balancing intuition with formal arguments.
Guidelines for paper presentations
If presenting your original work, which is great!, we would recommend shooting for a 40-minute talk, or a 20-minute talk if you are splitting the day with someone.
When you present a paper to the class, the goal is to get everyone on board with the basics but leave time for discussion. That usually means going over the motivation, basic definitions and setting, and overviewing the key results (perhaps relative to the existing literature). It would be great to get that far within the first 20 minutes. From there, we can open it up to discussion, and choose which parts to dive into.
As for the format, when in doubt, slides or a whiteboard presentation are great. Here you would prepare slides or whiteboard content on the portion described above. You could also add material for the open discussion portion, e.g. with slides/prepared material for a proof (sketch) of particular key results in case folks want to go through it. It is also fine to defer to the presenting directly from the paper at that point. In some cases, you could present entirely from the paper, but it's good to check with us (Bo and Raf) first.
Requirements
Each enrolled student is expected to present at least once to the group. In person attendance is important to the discussions, though we hope to offer a Zoom option for those unable to attend in person on any given day.