Course information

Syllabus

Note 1: the following table contains all contents but may be subject to change since I could be traveling to other conferences or events, and/or the guest lecturers could be changed due to their private matters. Please be aware of my emails.

Note 2: each class lasts 80 minutes = 55min lecture + 5min buffer + 20min presentation. Each student is supposed to finish one presentation.

Week & Date Topic/Content Material/Suggested Reading Note
1 (08/28) L01: Introduction
1. On the Opportunities and Risks of Foundation Models
  1. What is Generative AI and How Does It Work? (video) | | | 2 (09/02, 09/04) | L02: Autoencoders and VAE | 1. Auto-Encoding Variational Bayes
  2. Autoencoders
  3. Extracting and composing robust features with denoising autoencoders | Homework 1 starts | | 3 (09/09, 09/11) | L03: Generative Adversarial Nets | 1. Generative adversarial nets
  4. Unpaired image-to-image translation using cycle-consistent adversarial networks
  5. Conditional generative adversarial nets | | | 4 (09/16, 09/18) | L04: Diffusion Models | 1. Denoising Diffusion Probabilistic Models
  6. Variational diffusion models
  7. A survey on generative diffusion models | | | 5 (09/23, 09/25) | Guest Lecture on 09/23: Hao Chen, Google Deepmind: Diffusion models L05: Autoregressive Models (Slides) | 1. Attention is All you Need
  8. Slight Corruption in Pre-training Data Makes Better Diffusion Models | Homework 1 due 09/28 | | 6 (09/30, 10/02) | L06: Large Language Models | 1. CulturePark: Boosting Cross-cultural Understanding in Large Language Models
  9. Promptrobust: Towards Evaluating the Robustness of Large Language Models on Adversarial Prompts
  10. Constitutional AI: Harmlessness from AI Feedback | Homework 2 starts | | 7 (10/07, 10/09) | L07: AI Agents | 1. CompeteAI: Understanding the Competition Behaviors in Large Language Model-based Agents | 10/09—10/12 fall break | | 8 (10/14, 10/16) | L08: Pre-training | 1. SimCLR: A Simple Framework for Contrastive Learning of Visual Representations
  11. Masked autoencoders are scalable vision learners
  12. Momentum contrast for unsupervised visual representation learning | Homework 2 due | | 9 (10/21, 10/23) | L09: Post-training and Adaptation

| 1. Lora: Low-rank adaptation of large language models 2. On the effectiveness of parameter-efficient fine-tuning 3. Challenging big-bench tasks and whether chain-of-thought can solve them | Homework 3 starts | | 10 (10/28, 10/30) | Guest Lecture on 10/28: Sherry Wu, Carnegie Mellon University L10: Evaluation of GenAI | 1. Dyval: Dynamic evaluation of large language models for reasoning tasks 2. Promptbench: A unified library for evaluation of large language models | | | 11 (11/04, 11/06) | L10: Evaluation of GenAI (Continued) | 1. Promptrobust: Towards evaluating the robustness of large language models on adversarial prompts 2. Personalized Safety in LLMs: A Benchmark and A Planning-Based Agent Approach 3. Understanding and Mitigating the Label Noise in Pre-training on Downstream Tasks | 11/04 Election Day Homework 3 due | | 12 (11/11, 11/13) | L11: Safety and Robustness | 1. The good, the bad, and why: Unveiling emotions in generative ai 2. Culturellm: Incorporating cultural differences into large language models | Homework 4 starts | | 13 (11/18, 11/20) | Guest Lecture on 11/18: Damien Teney, Idiap Research Institute Guest Lecture on 11/20: Haohan Wang, UIUC | 1. Selective mixup helps with distribution shifts, but not (only) because of mixup 2. OOD-Chameleon: Is Algorithm Selection for OOD Generalization Learnable? | | | 14 (11/25, 11/27) | L12: Alignment and Human-AI Collaboration (Online) | 1. Fairness and Abstraction in Sociotechnical Systems 2. [Why Should I Trust You?": Explaining the Predictions of Any Classifier](https://dl.acm.org/doi/pdf/10.1145/2939672.2939778?) | 11/25 Recognition Day 11/26—11/30 Thanksgiving | | 15 (12/02, 12/04) | L13: Fairness and Interpretability (Recording) L14: GenAI Privacy (Online) | 1. Membership inference attacks against machine learning models 2. Extracting Training Data from Large Language Models 3. Regulating ChatGPT and Other Large Generative AI Models: A Privacy and Data Protection Perspective | Homework 4 due |

Grading

Acknowledgements