291K Machine Learning (Fall 2022)

Course Description

Machine Learning is about developing systems that automatically improve their performance through experience. It has found massive applications in real products. Examples include systems that recommend online videos, automatic translating languages, and autonomous driving vehicles. This advanced course gives a general and in-depth introduction to the theory, models, and practical algorithms for machine learning. Topics include learning problems (supervised learning, unsupervised learning, self-supervised learning, reinforcement learning, and online learning); modeling tools (graphical models, neural networks, kernel methods, tree methods); as well as theoretical foundations (learning theory, optimization). We focus on both the principles, analytical skills and implementation practice. This course is suitable for graduate students who want to pursue a career in AI/ML, to conduct research in this area, or apply ML methods in their own projects. No prior knowledge of machine learning is assumed.

Instructor

Lei Li (Office Hour: HFH2121, Thursdays 2-3pm, book a slot here or drop in)

Yu-Xiang Wang (Office Hour: HH 2013, Tuesdays 1-2 pm, or by appointment.)

Teaching Assistant

Xuandong Zhao (Office Hour: HH 2118, Mondays 4-5pm)

Time and Location

Tue/Thur 11am - 12:50pm (PHELP 3526)

Reference Textbook

[PRML] Pattern Recognition and Machine Learning, Chris Bishop. available online
[FML] Foundations of Machine Learning. Mehryar Mohri, Afshin Rostamizadeh, and Ameet Talwalkar. available online.
[D2L] Dive into Deep Learning. Aston Zhang, Zack Lipton, Mu Li, Alex Smola. available online.

The textbook below is a great resource for those hoping to brush up on the prerequisite mathematics background for this course.

[MML] Mathematics for Machine Learning, Marc Peter Deisenroth, A. Aldo Faisal, and Cheng Soon Ong. Free online.

Prerequisites

Linear algebra (MATH 3B), Vector Calculus (6A), Probability and Statistics (PSTAT 120A, 120B), algorithms (CS 130A & 130B), and familiarity with Python programming.

Homework Submission & Grading

We use Gradescope to grade all homework. You will be automatically added.
Late Days: 4 days in total for all assignments. No late day for project report.
Extra credit for lecture volunteer scribing. Latex Template

Discussion Forum

We will use Edstem platform. Please signup here

Policy

Please read the following Link carefully!

Syllabus

#	Date	Topic	Reading	Slides	Homework
1	9/22	Introduction, Spam filter	PPML 1.1, PRML 4.1.1, 4.1.2	lec1	HW0, data
2	9/27	Supervised Learning, Loss function, Model Selection	PRML 1.3, 3.1, 3.2	lec2, [annotated]	HW1 out
3	9/29	Unsupervised Learning, dimensionality reduction	PRML 9.1, 12.1	lec3	HW0 due
4	10/4	Optimization basic: Gradient Descent and SGD	FML 2.3, 11.2.1, MML 7.1	lec4, [annotated]
5	10/6	Feedforward Neural Networks	D2L 5	lec5	Proposal due
6	10/11	Convolutional Neural Networks	D2L 7 & 8	lec6	HW1 due, HW2 out
7	10/13	Sequence Modeling and Recurrent Neural Networks	D2L 9 & 10	lec7
8	10/18	Attention Mechanism and Transformers	D2L 11, 15.8, 15.9, 15.10	lec8
9	10/20	Graphical Models and MLE	PRML 8.1, 8.2	lec9
10	10/25	Gaussian Mixture Models, EM	PRML 9, 12.2,	lec10
11	10/27	Linear Dynamical Systems	PRML 13.3	lec11	HW2 due, HW3
12	11/1	Undirected Graphical Models, Conditional Random Fields	PRML 8.3, 8.4. CRF note	lec12
13	11/3	Deep Latent Models and Variational Inference	PRML 10, VAE, Sentence VAE	lec13
14	11/8	Sampling Methods	PRML 11	lec14	Midterm report due
15	11/10	Convex Optimization	FML Appendix B, FML 5.1 - 5.3	lec15, [annotated]
16	11/15	Duality and Support Vector Machine	FML 8.1,8.2	lec16, [annotated]
17	11/17	SVM (Part II) and Online Learning	FML 2.1-2.3, 3.1	>lec17, [annotated]	HW3 due
18	11/22	Online Learning (Part II ), Intro to RL	FML 17.1-17.4	lec18, [annotated]
19	11/29	Reinforcement learning (Part I)	Note 1 , Note 2 , and FML 17.5,	lec19, [annotated]
20	12/1	Reinforcement Learning (Part II)	Notes	lec20, [annotated]
	12/5	Final project poster presentation