CS499/599 | AI539 :: S26 :: Trustworthy Machine Learning



Textbooks

No required textbook. Reading materials will be provided on the course website and/or distributed in class. If you lack the basics in machine learning (or deep learning), the following bibles can be helpful:

  • [FOD'20] Mathematics for Machine Learning [PDF]
  • [B'06] Pattern Recognition and Machine Learning [PDF]
  • [GBC'16] Deep Learning [PDF]

Prerequisites

This course requires a basic understanding of ML:

  • CS 434: Machine Learning and Data Mining (required)
  • AI 535: Deep Learning (optional).

Grading

Your final grade for this course will be based on the following scheme:

  • 30%: Written paper critiques [Details]
  • 10%: In-class paper presentation [Details]
  • 20%: Homeworks (HW 1-4) [Details]
  • 30%: Group project [Details]
  • 10%: Final exam

  • Up to 10%: Extra point opportunities
    • +5%: Outstanding project work
    • +5%: Submitting the final report to workshops

Instructional Format


Latest Announcements


Schedule

Date Mode Topics Notice Readings
Part I: Overview and Motivation
Mon.
03/30
In-person Introduction
[Slides]
[HW 1 Out] (Classic) SoK: Security and Privacy in Machine Learning
Wed.
04/01
Online ML Overview Team-up [Sheet] (Book) Dive into Deep Learning
Part II: Adversarial Examples
Mon.
04/06
In-person Basics
[Slides]
- (Classic) Explaining and Harnessing Adversarial Examples
(Classic) Towards Deep Learning Models Resistant to Adversarial Attacks
Wed.
04/08
Online Black-box Attacks
[Slides]
- (Classic) Delving into Transferable Adversarial Examples and Black-box Attacks
(Classic) Prior Convictions
(Classic) Improving Black-box Adversarial Attacks with a Transfer-based Prior
Mon.
04/13
In-person Defenses
[Slides]
[HW 1 Due]
[HW 2 Out]
(Classic) Certified Adversarial Robustness via Randomized Smoothing
(Recent) (Certified!!) Adversarial Robustness for Free!
Wed.
04/15
Online Hands-on I
Checkpoint I Prep
Hands-on Lab: Emulating Adversarial Attacks on ML Models
Topic Introduction and Team Building
Mon.
04/20
In-person Group Project - Checkpoint Presentation 1
Wed.
04/22
No class - - Post-Checkpoint Presentation I :: HW2 Completion
Part III: Data Poisoning
Mon.
04/27
In-person Preliminaries
[Slides]
[HW 2 Due]
[HW 3 Out]
(Classic) Poisoning Attacks against SVMs
(Classic) Manipulating ML: Poisoning Attacks and Countermeasures...
Wed.
04/29
Online Attacks on NN
[Slides]
- (Classic) Poison Frogs! Targeted Clean-Label Poisoning Attacks on NNs
(Classic) MetaPoison: Practical General-purpose Clean-label Data Poisoning
Mon.
05/04
In-person Defenses
[Slides]
- (Classic) Certified Defenses for Data Poisoning Attacks
(Classic) Data Poisoning against DP Learners: Attacks and Defenses
Wed.
05/06
Online Hands-on II
Checkpoint II Prep
- Hands-on Lab: Emulating Poisoning Attack on ML Models
Mon.
05/11
In-person Group Project - Checkpoint Presentation 2
Wed.
05/13
No class - [HW 3 Due]
[HW 4 Out]
Post-Checkpoint Presentation II :: HW3 Completion
Part IV: Privacy
Mon.
05/18
Online(Async) S&P Travel - (Recorded materials on Canvas)
(Classic) Privacy Risk in ML: Analyzing the Connection to Overfitting
(Recent) Membership Inference Attacks From First Principles
Wed.
05/20
Online(Async) S&P Travel - (Recorded materials on Canvas)
(Classic) Evaluating and Testing Unintended Memorization in NNs
(Recent) Extracting Training Data from Large Language Models
Mon.
05/25
No class - Memorial Day
Wed.
05/27
In-person Defense
[Slides]
- (Classic) Deep Learning with Differential Privacy
(Recent) Evaluating DP ML in Practice
Mon.
06/01
Online Hands-on III
Final Presentation Prep
- Hands-on Lab: Emulating Membership Inference on ML Models
Wed.
06/03
In-person Group Project - Final Project Presentations (Showcases)
Finals Week
Mon.
06/08
No class Final Exam - Final Exam & Submit Your Final Project Report.
Wed.
06/10
No class Final Exam [HW 4 Due] Late Submissions for HW 1-4.