Generative Data Science

STATS C263 - Generative Data Science

<p>Lecture, three hours; discussion, one hour. Enforced requisite: one course from course C161, Computer Science M146, Electrical and Computer Engineering M146, C147, or Mathematics 156. Designed for students interested in generative data science, an emerging field in the intersection of generative artificial intelligence (AI), data science, and machine learning. Introduction to core concepts of generative data science, covering generation, evaluation, and utilization of generative data such as image, text, conversation, and tables. Focus on the evaluation of privacy-preserving effect of generative data. Concurrently scheduled with course C263. P/NP or letter grading.</p>

Statistics

College of Letters and Science

Container 1

Statistics and Data Science

STATS C163 - Generative Data Science

Master generative AI algorithms such as diffusion models, with a focus on tabular generative models (e.g., CTGAN, AIM, TabDDPM

Develop a deep understanding of data privacy, including differential privacy and privacy attacks, and learn to assess and mitigate privacy risks in the use of generative data

Apply generative models to digital marketing tasks like click-through rate prediction, and empirically evaluate fidelity, utility, and privacy of generative data in this context

General Catalog

Generative Data Science

STATS C163

4 units