Data-Centric AI. 3 credits, 3 contact hours

Catalog description

Prerequisites: IS 465 or CS 301 or DS 340 . Data-centric AI focuses on the systematic design of data to improve machine learning outcomes, rather than prioritizing the design and optimization of model architectures and their parameters (model-centric AI). In this course, students will learn to enhance data quality, consistency, and relevance to boost model performance—essential skills for real-world AI applications. Course topics may include AI task design and data requirements, data acquisition, data cleaning and quality assessment, data annotation, annotator reliability, active learning, programmatic labeling, confident learning, data augmentation, data synthesis, data balancing, and data monitoring. By the end of the course, students will be able to use modern tools and frameworks to engineer effective datasets and multi-stage training pipelines to improve AI systems.