Machine Learning (CSCI-GA.2565-001)

Course Description

The course covers general machine learning methods including generalized linear models, graphical models, causal inference, and reinforcement learning. There will be homework assignments along with a final project. Reading responses will be required every week along with a scribe of one lecture. The class will be live over zoom.

Lecture Location

Tuesday 7:10-9:00pm; Location: SILV 405 (100 Wash. Sq. East)

Reading Assignments

Each week, you will study from an assigned reading in the days leading up to lecture. These readings are mostly from the two common texts on ML, Bishop and Murphy, along with the Deep Learning book and some articles.

You should write a small (1/3-1/2 page) reflection on the reading. You can summarize the reading in a sentence or two, but the main point is let you ask any remaining questions you have. In class, we will gladly address questions.

As with the homeworks, you should submit these on Gradescope by 5pm EST on the day of the lecture.

• Introduction and Overview (1/25/22)
• Supervised Learning (2/1/22)
  • [Bishop PRML, Chapter 3.1]
• A Bayesian Interlude (2/8/22)
  • [Bishop PRML, Chapter 3.2]
• Regularization + Generalized Linear Models (2/15/22)
  • [Murphy, Probabilistic Machine Learning: An Introduction, Chapter 12.1-12.3 new edition, Chapter 9.1-9.3 in 2012 edition]
• Neural Networks (2/22/22)
  • Deep Feedforward Networks
• Trees and Forests (3/1/22)
  • Random Forests
  • Gradient Boosting
• Latent Variable Models (3/8/22)
  • [Bishop PRML, Chapter 8.1-8.3]
• Computing with Latent Variable Models (3/22/22)
  • [Bishop PRML, Chapter 10.1-11.2]
  • Supplementary: Chapter 10, in Murphy Book 2
• Deep Generative Models (3/29/22)
  • Generative Adversarial Nets
  • WaveNet
  • Deep Latent Gaussian Model
  • Supplementary: Chapter 21-24 + 27, in Murphy Book 2
• Causal Inference (4/5/22)
  • Statistics and Causal Inference
  • [Morgan and Winship, Counterfactuals and Causal Inference: Methods And Principles For Social Research (Analytical Methods for Social Research), 2nd ed, Chapter 2]
  • Supplementary: Chapter 38, in Murphy Book 2
• Reinforcement Learning (4/12/22)
  • Sutton and Barto, Reinforcement Learning: An Introduction, Chapter 1 and Chapter 13
  • Human-level Control through Deep Reinforcement Learning
• Cautionary Tales in Machine Learning (4/19/22)
  • Hidden Stratification Causes Clinically Meaningful Failures in Machine Learning for Medical Imaging
• Concept Reviews (Victory Lap) (4/26/22)
• Class Presentations (5/3/22)

Assignments

The graded assignments in this class are the homeworks, reading responses (see above), and project. Finally, students will be required to scribe one lecture in latex.

Homeworks are due at 5pm EST on the lecture day they are due.

• Homework 0 Open date: 1/25/22 Due date: 2/8/22

• Homework 1 Open date: 2/8/22 Due date: 2/25/22

• Homework 2 Open date: 2/24/22 Due date: 3/10/22

• Homework 3 Open date: 3/8/22 Due date: 3/29/22

• Homework 4 Open date: 4/5/22 Due date: 4/26/22

Projects

There will be a research project due at the end of the term.

• Submit proposals 3/20/22

• Check-in with TAs: 3/20/22 - 4/12/22

• Project Presentations: 4/26/22 and 5/3/22

• Paper Submission: 5/10/22

Late Policy

We allow you to spend 9 late days without penalty throughout the semester. No need to email us when using them. A whole late day is used for submissions past ~11:59pm EST.

Office Hours

All office hours are online and in EST time. You can find links on Brightspace.

• Monday 5-6pm, Mark Goldstein
• Tuesday 5:30-6:30pm, Rajesh Ranganath
• Thursday 10-11am, online, Mark Goldstein
• Friday 11am-12noon, Anugya Srivastava