Topics in Machine Learning: Compositional AI Systems
COMPSCI 2821R
Subject & Catalog Number
Course Information
Description
This course is a seminar-style course where students will help present papers every other day. We'll survey different compositional AI systems, ranging from classical AI systems such as logic, constraint satisfaction, production systems, to modern compositional AI systems such as multi-agent language models, neural-symbolic systems, and compositional generative models. We'll discuss the unified design choices across each compositional system, as well as tradeoffs between different compositional systems. Finally, we'll talk about how such compositional systems can be applied in diverse applications such as robotics, natural language processing, reasoning, and general artificial intelligence.
Course Notes
This course was previously numbered CS 282BR.
Available for Harvard Cross Registration
