Lectures
| Lecturer | Varun Kanade |
| Hours | Mondays, Tuesdays and Thursdays 16h-17h |
| Location | Lecture Theatre A (Wolfson Building) |
| Notes | (Draft Lecture Notes) |
Note 1. A set of lecture notes (LN) covering the entire course will be posted. These are not meant to be used as a substitute for reading the recommended textbooks and other posted reading. There might be some delay in updating the lecture notes. Should you wish to read up ahead of lectures, you may follow the material posted on the Computational Learning Theory course from Michaelmas Term 2018. Please make sure to make use of the most updated material when revising.
Note 2. There will be parts of suggested reading material that you struggle with or may seem not directly related to what we covered in the lecture. I suggest you skip these sections and only return to them once you've understood the material covered in the lecture and want to increase your understanding further.
Note 3. I expect the total number of lecture hours to be about 20. Lectures are scheduled for 24 hours to make it convenient to adjust the timetable.
Lecture Schedule
| Date | Topic | Reading |
|---|---|---|
| 14/10/2019 | Setting up the PAC Learning Framework |
Chapter 1.1 (LN) Chapter 1.1 (KV) |
| 15/10/2019 | PAC Learning Framework |
Chapter 1.2 (LN) Chapter 1.2 (KV) |
| 17/10/2019 | Intractability of Learning 3-term DNF; PAC Learning |
Chapter 1.3-8 (LN) Chapter 1.3-6 (KV) Valiant's paper |
| 21/10/2019 | Intractability of Learning 3-term DNF; Occam's Razor |
Chapter 1.6-8, 2 (LN) Chapter 1.5-6, 2 (KV) |
| 22/10/2019 | Occam's Razor Theorm; VC Dimension |
Chapter 2 (LN) Chapter 2, 3 (KV) |
| 24/10/2019 | VC Dimension: Sample Complexity Upper Bounds |
Chapter 3 (LN) Chapter 2, 3 (KV) |
| 28/10/2019 | VC Dimension: Sample Complexity Lower Bounds |
Chapter 3 (LN) Chapter 3 (KV) |
| 29/10/2019 | Boosting; Adaboost |
Chapter 4 (LN) Adaboost survey Adaboost article |
| 31/10/2019 | Cryptographic Hardness of Learning |
Chapter 5 (LN) Chapter 6 (KV) |
| 4/11/2019 | Cryptographic Hardness of Learning; MQ Model |
Chapter 5, 6 (LN) Chapter 8 (KV) |
| 5/11/2019 | Learning DFAs using MQs + EQs |
Chapter 8 (KV) |
| 7/11/2019 | Learning MONOTONE-DNF using MQs + EQs Learning in the presence of noise |
Chapter 5 (KV) |
| 11/11/2019 | SQ & Learning in the presence of noise |
Chapter 5 (KV) |
| 12/11/2019 | Learning PARITIES is hard in SQ |
Chapter 5 (KV) |
| 14/11/2019 | Learning Real-Valued Functions; Convex Optimization |
Chapter 10 (MRT) |
| 18/11/2019 | Learning GLMs |
Chapter 10 (MRT) |
| 19/11/2019 | Rademacher Complexity and Generalization |
Chapter 3 (MRT) |
| 21/11/2019 | Online Learning; Perceptron |
Chapter 8 (MRT) |
| 25/11/2019 | Winnow & Learning with expert advice |
Chapter 8 (MRT) |
| 26/11/2019 | Learning Decision Trees using MQs (not examinable) |
[Ben Worrell's notes] |
| 28/11/2019 | Learning Decision Trees using MQs (not examinable) |
|
| 2/12/2019 | Learning with expert advice; Multiplicative Weight Updates |
Chapter 8 (MRT) |