This course is based on in-person lectures, in-person recitation sessions, class notes, and homework.
The class notes in the Notes column below are the official study materials. All appendices in the notes are optional. The note set n_00 next to Overview and Logistics covers part of the prerequisites. You are supposed to know that material already, so this set is optional reading on case you need a refresher.
Lecture slides are also made available for your convenience in the Slides column below. However, they are optional reading and should not be used as your main study materials.
Lectures are automatically recorded on the Panopto system, and are available through Sakai. You need a Duke affiliation to access the recordings.
Prerecorded videos are also available as a backup in the Videos column below for most but not all topics, in case you miss a class or want to review something. However, these videos are optional, and they cover only approximations of the official material.
| # | Topic | Notes | Slides | Prerecorded Videos | HW Due |
|---|---|---|---|---|---|
| 0 | Overview and Logistics | (n_00) | s_00 | Sep 7 | |
| 1 | Functions and Data Fitting | n_01 | s_01 | v_01_1, v_01_2 | |
| 2 | Machine Learning Concepts | n_02 | s_02 | v_02 | |
| 3 | Nearest Neighbor Predictors | n_03 | s_03 | v_03 | |
| 4 | Function Optimization | n_04 | s_04 | v_04 | |
| 5 | Linear Predictors Part 1 | n_05 | s_05 | v_05 | |
| 6 | Linear Predictors Part 2 | n_06 | s_06 | v_06 | |
| 7 | Validation and Testing | n_07 | s_07 | v_07 |
| October 26: Midterm Exam on topics 1-5 (not 6) |
| 8 | Support Vector Machines | n_08 | s_08 | — | |
| 9 | Kernels | n_09 | s_09 | — | |
| 10 | Decision Trees and Forests | n_10 | s_10 | v_10_1, v_10_2 | |
| 11 | Neural Networks | n_11 | s_11 | v_11 | |
| 12 | Training Neural Networks | n_12 | s_12 | v_12 | |
| 13 | Improving Generalization | n_13 | s_13 | v_13 | |
| 14 | Topics not Covered | -- | s_14 | -- |
| December 16, 9am: Final Exam on topics 6-12 (13, 14 not on the exam) |
COMPSCI 371, Duke University, Site based on the fluid 960 grid system