This page will house information about the midterm. Full information about the midterm, along with what to study, how to study, and a set of practice problems I will be updating is all included in this file: Midterm Review Sheet. This sheet will be continually updated, so please check back here for problems for later lectures! Currently, the problems are up to:

  • Lecture 1: Statistical Learning Framework
  • Lecture 2: Optimization and Gradient Descent
  • Lecture 3: Regularization and Loss Functions
  • Lecture 4: Convex Optimization and SVM
  • Lecture 5: Features and Kernels

Solutions to these problems will be posted here: Midterm Solutions.

Basic Information. The midterm for this class will take place on March 10, 2026 2:45pm - 4:45pm during the usual class time, in the usual place in lieu of the lecture. Please arrive early if you can, because we will be starting at 2:45pm sharp. You will be allowed the following materials:

  • A pen or pencil. I will provide sheets of scrap paper during the midterm.
  • One-double sided 8.5 × 11 sheet of paper containing whatever materials you like. This can be handwritten or typed and there is no restriction on what this sheet contains.

Any collaboration between students or with outside sources via phones, laptops, “smart glasses,” smoke signals, pagers, carrier pigeons, etc. is strictly prohibited.

As we have stated from the first lecture, the course policy is that students who cannot make the midterm will need to take a grade of “Incomplete” for the course. If you believe that you will not be able to make it for whatever reason, please reach out to the instructors as soon as you can so we can discuss what this means!

Midterm Format. You can be assured that the format of the midterm will be as follows:

  • One section for each of the first six lectures of the course (up to and including MLE & Conditional Probability Models).
  • Each section except the MLE & Conditional Probability Models section will have exactly two parts: three True/False questions and one multi-part short answer problem, worth a total of 18 points.
  • True/False. Each True/False problem will be worth 3 points each. You will need to answer whether a statement is true or false with a one-sentence justification. 1 point is awarded for the correct True/False without justification, but the full 3 points will only be awarded for a correct justification.
  • Short answer. Each short answer problem is worth 9 total points. The short answer problems will involve working through a short toy machine learning problem, doing some short derivations, or answering conceptual questions. The samples in the above document should be a good representation of their difficulty level.
  • The MLE & Conditional Probability Models section of the exam will only have a single short answer problem worth 9 points.