ECE693B: Information Theoretic Methods in Generative AI

Meeting Days/Time: Monday & Wednesday, 3:30 PM – 4:45 PM
Location: KUY 207
Semester Dates: August 24, 2026 – December 18, 2026
Instructor: Haoyue Tang — Department of Electrical and Computer Engineering, University of Hawaiʻi at Mānoa
Email: haoyue@hawaii.edu
Office Hours: Friday 10:00–11:00 AM, POST 205E

[Course Syllabus]

Outline

This course provides an introduction to deep generative models, the information-theoretic principles that underlie them, and their applications in image generation and feature learning. Students will study the theory and practice behind the major families of generative models — autoregressive models, variational autoencoders (VAEs), normalizing flows, flow matching, generative adversarial networks (GANs), energy-based models, and score-based/diffusion models — and how they evolve towards today's generative AI systems. The course combines lectures, written & coding assignments, a reading report, and a final project.

Prerequisites

Basic knowledge of probability and machine learning. Ability to code in PyTorch (use of AI coding assistants such as Claude or GPT is permitted).

Grading

• Assignments (3 × 15%): 45%

• Literature Reading Report: 20%

• Final Project: 35%

Course Policies

• Late Work: Late submissions will be graded at 50% of the original score.

• Academic Integrity: Students are expected to adhere to the UH Mānoa Student Conduct Code.

• Accessibility (KOKUA): Students needing accommodations should contact the KOKUA Program.

General References

No required textbook. Readings and lecture materials will draw on publicly available resources:

• Stanford CS236: Deep Generative Models (Stefano Ermon)

• Additional readings will be posted per lecture on Lamaku.

Schedule

Topic sequence follows the general arrangement of Stanford CS236. University holidays are marked below. This schedule is a draft and may be adjusted as the semester progresses.