This course provides academic credit for independent thesis research conducted under the supervision of a Georgia Tech faculty advisor. The course does not involve regular class meetings, assignments, or examinations. The scope and direction of research are determined by the student in consultation with the thesis advisor, consistent with the requirements of the degree program.

In the AI era, two major trends are reshaping computer graphics: (1) visual content is becoming increasingly realistic through data-driven techniques and neural rendering frameworks like Neural Radiance Fields and Gaussian splatting, strengthening the link between virtual models and real-world data; and (2) visual content creation pipelines are becoming more accessible, allowing creators to use diverse input methods, such as Unreal Engine, image and video examples, or natural language, to generate realistic content at interactive or real-time rates. This shift marks a move away from traditional programmable shader development and hand-crafted 3D assets toward end-to-end differentiable and AI-model-driven generation.

The CGAI course offers a comprehensive introduction to modern computer graphics, emphasizing AI-powered advancements in modeling, rendering, simulation, and animation. Students will explore key topics such as Neural Signed Distance Fields (SDF), Neural Radiance Fields (NeRF), 3D Gaussian Splatting (3D-GS), and Position-Based Dynamics (PBD), alongside advanced areas like generative 3D graphics using diffusion models and LLM-driven pipelines. The course also connects to traditional graphics topics like BRDF rendering, point-based rendering, and physically-based animation. Positioned as an advanced follow-up to CS3451 Computer Graphics, CGAI equips students with cutting-edge tools and a forward-looking mindset to engage with the evolving landscape of AI-integrated graphics applications.

Introduction to computing principles and programming practices with an emphasis on the design, construction and implementation of problem solutions use of software tools.

Introduction to computing principles and programming practices with an emphasis on the design, construction and implementation of problem solutions use of software tools.

Small-group or individual investigation of advanced topics in computing. Guided study and research.

An investigation of significant areas of information and computer science. Guided study and research.

Group discussion of advanced topics in information and computer science. May not be used by computer science majors for degree credit.

This course develops a principled, unified foundation for representation learning  through the lens of spectral decomposition. Starting from the sufficiency of spectral representations for downstream tasks, we systematically build a framework that connects classical component analysis methods (PCA, CCA, Laplacian Embedding) to modern self-supervised learning algorithms (SimCLR, BYOL, CLIP, DINO) and extends to applications in reinforcement learning, causal inference, and controllable generation.

    The course follows the theoretical backbone of "Spectral Ghost in Representation Learning" (Dai, Li & Schuurmans, 2026), supplemented by the RL-specific treatment in "Spectral Representation-based Reinforcement Learning" (Gao, Sun, Li, Schuurmans &  Dai, 2026).