This course is CMPT 479/982: AI in Security. Welcome!

This is a seminar-based course. Our objective is to conduct original research. To learn to do so, we will read, discuss, and critique academic papers. We will focus on papers that use machine learning in security & privacy research.

The course will be organized as follows:

• Phase 1: We will start with around 4 weeks of lectures. The lectures will cover two major topics: Machine Learning and Security. Because these materials are significantly condensed - we are turning full courses into 1-2 weeks of material! - you are expected to pursue self-learning.
• Phase 2: We will discuss and critique papers for the rest of the semester, around 8 weeks. Finally, we will have 1 week of presentations on your projects.
  • There will be 3 papers every week, with 25 minutes for the presentation followed by 20 minutes of discussions.

Classes are held in RCB 5120 on Tuesday 11:30AM - 2:20 PM.

Recordings will be available from Information Systems; the link will be posted here.

Please e-mail me if you need to meet me at other times or in person.

Grading

Your mark will consist of the following:

• Participation: 20%
  • Participation grade is given for participating in in-class discussions during Phase 2.
• Assignments: 20%
  • There will be 2 assignments that will guide you on using machine learning in security & privacy research.
• Presentation: 10%
  • Present one (or more) academic papers in Phase 2. Your highest grade will be taken.
• Project: 30%
  • Conduct original research and demonstrate your work. Group sizes up to 4.
• Oral Exam: 20%
  • We will have an oral exam in the exam period. We will talk about your project and the course material.

Detailed rubrics are here.

Deadlines

Choose paper(s) to present: May 29. https://grow.withlome.com/a/8a263f80-dc8c-418e-8d2e-4c90d9ca8ca4

Form project group in CourSys: Jun 5. Please create your group in CourSys and give your group a recognizable unique name.

Assignment 1: Jun 15.

Assignment 2: Jul 13.

Project submission: Aug 13.

Final Exam (Oral Exam): Aug 17 and 18. In my office, TASC1 9015.

Slides (Phase 1)

Introduction

Machine Learning

Machine Learning - Neural Networks

Security

Notes on computer security research

Papers (Phase 2)

More papers will be added.

Jun 16 - Anomaly Detection - Malware Detection

A Unifying Review of Deep and Shallow Anomaly Detection. Ruff et al. Proceedings of the IEEE, 2021.

ADBench: Anomaly Detection Benchmark. Han et al. NeurIPS, 2022.

Prudent Practices for Designing Malware Experiments: Status Quo and Outlook. Rossow et al. IEEE S&P, 2012.

Malware Detection by Eating a Whole EXE. Raff et al. AAAI Workshop on Artificial Intelligence for Cyber Security, 2018.

Jun 23 - Anomaly Detection - Intrusion Detection

Toward a reliable anomaly-based intrusion detection in real-world environments. Viegas et al. Computer Networks, 2017.

MalBoT-DRL: Malware Botnet Detection Using Deep Reinforcement Learning in IoT Networks. Al-Fawa'reh et al. IEEE IoT Journal, 2024.

"Why 6?" Defining the operational limits of STIDE, an anomaly-based intrusion detector. Ran and Maxion. IEEE S&P, 2002.

Jun 30 - Model Security

One pixel attack for fooling deep neural networks. Su et al. IEEE Transactions on Evolutionary Computation, 2019.

Do not get fooled: Defense against the one-pixel attack to protect IoT-enabled Deep Learning systems. Husno and Anwar. Ad Hoc Networks, 2021.

Transferable Adversarial Perturbations. Zhou et al. ECCV, 2018.

Making Convolutional Networks Shift-Invariant Again. Zhang. ICML, 2019.

Jul 7 - Biometric Systems

Accessorize to a Crime: Real and Stealthy Attacks on State-of-the-Art Face Recognition. Sharif et al. CCS, 2016.

Going through the motions: AR/VR keylogging from user head motions. Slocum et al. USENIX, 2023.

Deep Learning Approach for Multimodal Biometric Recognition System Based on Fusion of Iris, Face, and Finger Vein Traits. Alay and Al-Baity. Internet of Things for Sensors and Biosensors, 2019.

On the Practicality of Motion Based Keystroke Inference Attack. Cai and Chen. International Conference on Trust and Trustworthy Computing, 2016.

Jul 14 - Traffic Analysis

Fingerprinting Attack on Tor Anonymity using Deep Learning. Abe and Goto. APAN, 2016.

Walkie-Talkie: An Efficient Defense Against Passive Website Fingerprinting Attacks. Wang and Goldberg. USENIX, 2017.

Traffic Morphing: An Efficient Defense Against Statistical Traffic Analysis. Wright et al. NDSS, 2009.

Online Website Fingerprinting: Evaluating Website Fingerprinting Attacks on Tor in the Real World. Cherubin et al. USENIX, 2022.

Jul 21 - Data Privacy

Inverting Gradients - How easy is it to break privacy in federated learning?. Geiping et al. NeurIPS, 2020.

"I know what you did last summer": query logs and user privacy. Jones et al. CIKM, 2007.

Reconstructing Training Data from Model Gradient, Provably. Wang et al. PMLR, 2023.

Deep learning with differential privacy. Abadi et al. CCS, 2016.

Jul 28 - Science

Dos and Don’ts of Machine Learning in Computer Security. Arp et al. USENIX Security, 2022.

SoK: Science, security and the elusive goal of security as a scientific pursuit. Herley et al. IEEE S&P, 2017.

“Real Attackers Don’t Compute Gradients”: Bridging the Gap Between Adversarial ML Research and Practice. Apruzzese et al. SaTML, 2023.

"Get in Researchers; We’re Measuring Reproducibility": A Reproducibility Study of Machine Learning Papers in Tier 1 Security Conferences. Olszewski et al.CCS, 2023.

Week 9 - Presentations

Assignments

Assignments will be posted here when released.

Other resources