10-715 Fall 2019: Advanced Introduction to Machine Learning

The rapid improvement of sensory techniques and processor speed, and the availability of inexpensive massive digital storage, have led to a growing demand for systems that can automatically comprehend and mine massive and complex data from diverse sources. Machine Learning is becoming the primary mechanism by which information is extracted from Big Data, and a primary pillar that Artificial Intelligence is built upon. This course is designed for Ph.D. students whose primary field of study is machine learning, or who intend to make machine learning methodological research a main focus of their thesis. It will give students a thorough grounding in the algorithms, mathematics, theories, and insights needed to do in-depth research and applications in machine learning. The topics of this course will in part parallel those covered in the general graduate machine learning course (10-701), but with a greater emphasis on depth in theory and algorithms. The course will also include additional advanced topics such as fairness in machine learning.

Time and location: The class is scheduled for Monday and Wednesday 10.30am to 11.50am in MM A14 and Friday 10.30am to 11.50am also in MM A14. The Friday slot will usually be used for recitations, but will also often be used for make-up classes (due to any travels on Mon/Wed).

Units: 12

Instructor: Nihar Shah

Syllabus: The topics covered will be similar to those covered last year.

References:
[SB] Understanding Machine Learning: From Theory to Algorithms by Shai Shalev-Shwartz and Shai Ben-David (available online)

Course staff and contact details:
Nihar Shah: nihars at cs dot cmu dot edu Office hours by appointment (GHC 8211). To set up a meeting, please send Nihar an email with your availability, as well as the topics you would like to discuss (e.g., specific lectures or project content).

Charvi Rastogi: crastogi at andrew.cmu.edu Office hours Friday 2-3 pm in the common area outside GHC 8011
Leqi Liu: leqil at andrew.cmu.edu Office hours Tuesday 2-3pm in the common area outside GHC 8011
Juyong Kim: juyongk at cs.cmu.edu Office hours Thursday 3-4pm in the common area outside GHC 8011

Tentative schedule (subject to change) and references: Note that the lectures will be taught on the board, and hence there are no "slides''.

Date	Topic	References/Remarks
Aug 26	Introduction, logistics, terminology, basics	Slides from past offering (ignore "logistics"), SB chapter 2
Aug 28	Perceptrons: Hope, hopelessness, and hope again	SB chapter 9
Aug 30	Optimization for ML	Notes
Sept 2	No class (labor day)
Sept 4	Class canceled due to flooding of the classroom	No, this is not a prank
Sept 9	Support vector machines	SB chapter 15
Sept 11	Kernel methods I	SB chapter 16
Sept 13	Recitation: Optimization
Sept 16	Kernel methods II	SB chapter 16
Sept 18	Decision trees, random forests, crossvalidation, bagging, bootstrapping, interpretability, explanability	SB chapter 18, Paper
Sept 20	Recitation: Tail bounds
Sept 23	(continued) Decision trees, random forests, crossvalidation, bagging, bootstrapping, interpretability, explanability Learning theory I	SB Chapters 2 - 5
Sept 25	Learning theory II	SB Chapters 2 - 6
Sept 30	Learning theory III	SB Chapters 2 - 6
Oct 2	Learning theory IV	SB Chapters 2 - 6
Oct 7	No class	There was a make-up class on Aug 30
Oct 9	Midterm (in class)	All material covered until Oct 4
Oct 14	Learning theory V, Interpolation regime	SB Chapters 6 - 7, Paper
Oct 16	Neural networks I	SB Chapter 20
Oct 21	Neural networks II	SB Chapter 20, Paper
Oct 23	Unsupervised learning: Clustering	SB Chapter 22
Oct 28	Guest lecture: Yuanzhi Li "Separation between Neural networks and Kernels"
Nov 1	Guest lecture: Andrej Risteski "Better understanding of modern paradigms in generative models"
Nov 4	Dimensionality reduction	SB Chapter 23
Nov 6	Boosting	SB Chapter 10
Nov 8	Recitation: MLE and MAP
Nov 11	Online learning	SB Chapter 21
Nov 13	Semi-supervised learning, Active learning, Multi-armed bandits	Transductive SVM, Active learning, Multi-armed bandits, Ranking via MABs
Nov 15	Recitation: Linear regression
Nov 18	Reinforcement learning	Survey
Nov 20	Graphical models	Graphical models
Nov 22	Recitation: Rademacher Complexity
Nov 25	Fairness	Hiring example, Paper 1, Paper 2, In peer review
Nov 27	No class (Thanksgiving)
Dec 2	Project presentation (in class)
Dec 4	Project presentation (in class)

Evaluation: Major homeworks (30%), mini homeworks (10%), midterm (25%), final project (25%), class participation (10%).
To audit the course, the student must do all of the above except the final project and get a passing grade.

Project: One of the course requirements is to do a project in a team of 3 or 4. Your class project is an opportunity for you to explore an interesting machine learning problem of your choice, whether empirically or theoretically. You may conduct an experiment (pick some datasets, and apply an existing or new machine learning algorithm) or work on a theoretical question in machine learning. The timeline for projects is as follows:

Homeworks Homeworks will be released on Diderot

Collaboration policy for homeworks

• You may discuss the questions
• Each student writes their own answers
• Each student must write their own code for the programming part
• Please don’t search for answers on the web, previous years’ homeworks, etc. (please ask the TAs if you are not sure if you can use a particular reference)