🤖 Artificial Intelligence Week 1 of 14 BSc · Y3 ⏱ ~50 min

Week 1: VAEs, GANs & Latent Space Representations

Master generative modeling: Variational Autoencoders, Generative Adversarial Networks, Diffusion Models, and Flow-based models — theory and PyTorch implementation.

University of America

AI303 — Lecture 1 · BSc Y3

🎬 CC Licensed Lecture

0:00 / —:—— 📺 MIT OpenCourseWare (CC BY-NC-SA)

🎯 Learning Objectives

Derive the VAE evidence lower bound (ELBO)
Train a GAN and understand mode collapse
Implement a DDPM diffusion model
Visualize and interpolate in latent spaces

Topics Covered This Lecture

Autoencoders & Latent Representations

Variational Autoencoders (VAE) & ELBO

Generative Adversarial Networks (GAN)

Diffusion Models: DDPM & Score Matching

📖 Lecture Overview

This first lecture establishes the foundational framework for Generative Models. By the end of this session, you will have the conceptual grounding and practical starting point needed for the rest of the course.

        Why this matters
        Master generative modeling: Variational Autoencoders, Generative Adversarial Networks, Diffusion Models, and Flow-based models — theory and PyTorch implementation. This lecture sets up everything that follows — make sure you understand the core concepts before proceeding to Week 2.
      

Key Concepts

The lecture introduces the four main pillars of this course: Autoencoders & Latent Representations, Variational Autoencoders (VAE) & ELBO, Generative Adversarial Networks (GAN), Diffusion Models: DDPM & Score Matching. Each will be explored in depth over the 14-week curriculum, with hands-on projects reinforcing theory at every stage.

# Quick Start: verify your environment is ready for AI303
import sys
print(f"Python {sys.version}")

# Check key libraries are installed
try:
    import numpy, pandas, matplotlib
    print("✅ Core libraries ready")
except ImportError as e:
    print(f"❌ Missing: {e} — run: pip install numpy pandas matplotlib")

This Week's Focus

Focus on mastering: Autoencoders & Latent Representations and Variational Autoencoders (VAE) & ELBO. These are the prerequisites for everything in Week 2. The concepts build on each other — do not skip the practice exercises.

📋 Project 1 of 3 50% of Final Grade

AI303 Project 1: Conditional Image Generator

Build a conditional GAN (cGAN) or conditional VAE that generates images conditioned on class labels. Train on CIFAR-10 or CelebA. Evaluate with FID and Inception Score.

cGAN or cVAE PyTorch implementation
Latent space visualization (t-SNE / UMAP)
FID score and Inception Score evaluation
Interpolation and conditional generation samples

50%

3 Projects

20%

Midterm Exam

30%

Final Exam

📝 Sample Exam Questions

These represent the style and difficulty of questions you'll see on the midterm and final. Start thinking about them now.

Conceptual Short Answer

Derive the ELBO for a VAE. What does each term (reconstruction + KL) represent?

Analysis Short Answer

What is mode collapse in GANs? Describe two training techniques that mitigate it.

Applied Code / Proof

Explain the score function in score-based diffusion models. How does it relate to DDPM?