Homework & Labs

Weight: 35% of the course grade (9 assignments).

Programming assignments in Python using TensorFlow or PyTorch. Assignments are cumulative and scaffold toward the capstone; case studies and security audits are embedded within the coding projects. The sequence follows the three pillars: AI for Security detectors first (HW1–HW4), then Security of AI attacks and defenses (HW5–HW8), then Trustworthy AI auditing and privacy (HW8–HW9). Each is graded on correctness of implementation, code quality, documentation, and adherence to the code submission standards (GitHub with README, inline comments, reproducibility, requirements.txt/environment.yml, random seeds).

Late policy: Late work is penalized 10% per day for up to 48 hours; work submitted more than 48 hours late receives no credit unless prior arrangements are made.

# Pillar Title Out Due Handout
HW1 AI for Security ML-for-Security Foundations and a Reproducible Baseline Detector Week 1 (Wed) Week 2 (Wed) README
HW2 AI for Security Network Intrusion & Anomaly Detection Week 2 Week 3 (Wed) README
HW3 AI for Security Malware Detection & Classification Week 3 Week 4 (Wed) README
HW4 AI for Security Phishing/Fraud Detection and AI for Security Operations Week 4 Week 5 (Wed) README
HW5 Security of AI Adversarial Attacks on Your Own Detector (FGSM and PGD, L∞) Week 5 Week 6 (Wed) README
HW6 Security of AI Data Poisoning and Backdoor Attacks (with a Simple Defense) Week 6 Week 7 (Wed) README
HW7 Security of AI Adversarial Defenses & Robustness Week 9 (Mon) Week 10 (Wed) README
HW8 Trustworthy AI Interpretability + Fairness Audit Week 10 Week 11 (Wed) README
HW9 Trustworthy AI Privacy-Preserving ML: DP-SGD & Membership Inference Week 14 (Mon) Week 15 (Wed) README

Out/Due weeks reflect the schedule; confirm exact dates on Canvas.

Quizzes & Participation (10%)

Short, low-stakes in-class or Canvas quizzes plus qualitative participation. Templates and rubric: