CMSC 726: Machine Learning

Welcome to the CMSC 726 course webpage for Fall 2018.

CMSC 726 is currently being taught by Soheil Feizi.

Announcements

• Review Probability, Linear Algebra and Convex Analysis.

• Lecture 1 (8/28): Basic Concepts + Linear Regression

  • Scribed notes for lecture 1 were sent out (request via email if you did not get it).

  • Reading: Section 9.2 of the text book + pages 1-3, 8-11 of notes

• Lecture 2 (8/30): Linear Algebra Review + Gradient Descent

  • Scribed notes for lecture 2 were sent out (request via email if you did not get it).

  • Reading: pages 4-7 of notes

  • Sign-up on course Piazza

  • Linear Regression Code

• Lecture 3 (9/4): Gradient Descent Convergence + Maximum Likelihood Estimation

  • Scribed notes for lecture 3 were sent out (request via email if you did not get it).

  • Reading: Section 14.1 of the text book+ Section 3 of notes

• Lecture 4 (9/6): Logistic Regression

  • Scribed notes for lecture 4 were sent out (request via email if you did not get it).

  • Reading: Section 9.3 of the text book+ part II of notes

  • Logistic Regression Code

• Lecture 5 (9/11): Convex Optimization, Lagrange Multipliers, KKT

  • Scribed notes for lecture 5 were sent out (request via email if you did not get it).

  • Reading: Section 5 of notes

• Lecture 6 (9/13): Support Vector Machines + Hinge Loss

  • Scribed notes for lecture 6 were sent out (request via email if you did not get it).

  • Reading: Part V of notes

• Lecture 7 (9/18): Kernels

  • Scribed notes for lecture 7 were sent out (request via email if you did not get it).

  • Reading: Section 16 of the text book + Section 7 of notes

• Lecture 8 (9/20): Coordinate Block Descent + Project Discussion

  • Scribed notes for lecture 8 were sent out (request via email if you did not get it).

  • Reading: Section 9 of notes

• Lecture 9 (9/25): Project Discussion

  • Scribed notes for lecture 9 were sent out (request via email if you did not get it).

  • Slides for the project discussion were sent out (request via email if you did not get it).

• Lecture 10 (9/27): PAC Learning + Uniform Convergence

  • Scribed notes for lecture 10 were sent out (request via email if you did not get it).

  • Reading: Chapters 3 and 4 from the text book.

• Lecture 11 (10/2): VC Dimension

  • Scribed notes for lecture 11 were sent out (request via email if you did not get it).

  • Reading: Chapters 5 and 6 from the text book.

• Lecture 12 (10/4): Rademacher Complexity

  • Scribed notes for lecture 12 were sent out (request via email if you did not get it).

  • Reading: Chapter 26 from the text book.

• Lecture 13 (10/9): Rademacher Calculus + Neural Networks

  • Scribed notes for lecture 13 were sent out (request via email if you did not get it).

  • Reading: Section 5 of notes

• Lecture 14 (10/11): Deep Learning + Non-convex Optimization

  • Scribed notes for lecture 14 were sent out (request via email if you did not get it).

  • Reading: Section 20 from the text book.

• Lecture 15 (10/16): Backpropagation + Higher Order Derivatives

  • Scribed notes for lecture 15 were sent out (request via email if you did not get it).

  • Reading: Section 5 of notes.

• Lecture 16 (10/18): Midterm Exam

  • Midterm solutions have been posted (request via email if you did not get it).

• Lecture 17 (10/23): Unsupervised Learning, Clustering

  • Scribed notes for lecture 17 were sent out (request via email if you did not get it).

  • Reading: Section 5 of notes.

• Lecture 18 (10/25): Expectation-Maximization

  • Scribed notes for lecture 18 were sent out (request via email if you did not get it).

  • Reading: notes and notes.

• Lecture 19 (10/30): KL Divergence + Variational AutoEncoders (VAEs)

  • Scribed notes for lecture 19 were sent out (request via email if you did not get it).

  • Reading: VAE Tutorial

• Lecture 20 (11/1): Variational AutoEncoders (VAEs)

  • Scribed notes for lecture 20 were sent out (request via email if you did not get it).

  • Reading: VAE Tutorial

• Lecture 21 (11/6): Principal Component Analysis (PCA)

  • Scribed notes for lecture 21 were sent out (request via email if you did not get it).

  • Reading: notes.

• Lecture 22 (11/8): Principal Component Analysis (PCA)

  • Scribed notes for lecture 22 were sent out (request via email if you did not get it).

  • Reading: notes.

• Lecture 23 (11/13): Kernel PCA + AutoEncoders

  • Scribed notes for lecture 23 were sent out (request via email if you did not get it).

  • Reading: KPCA paper.

• Lecture 24 (11/15): Reinforcement Learning + Finite Markov Decision Process

  • Scribed notes for lecture 24 were sent out (request via email if you did not get it).

  • Reading: Chapter 3 of book.

• Lecture 25 (11/20): Bellman Optimality + Policy Evaluation

  • Scribed notes for lecture 25 were sent out (request via email if you did not get it).

  • Reading: Chapter 4 of book.

• Lecture 26 (11/27): Value Iteration + Policy Iteration

  • Scribed notes for lecture 26 were sent out (request via email if you did not get it).

  • Reading: Chapter 4 of book.

• Lecture 27 (11/29): Finite Horizon MDPs + LQR setting

  • Scribed notes for lecture 27 were sent out (request via email if you did not get it).

  • Reading: Sections 1 and 2 of notes.

• Lecture 28 (12/4): Final Project Presentations

• Lecture 29 (12/6): Reinforcement Learning Examples

  • Scribed notes for lecture 29 were sent out (request via email if you did not get it).