Abstract

Note this seminar has been rescheduled from its original date and will be taking place at 4 pm.

Control theory is fundamental in the design and understanding of many natural and engineered systems, from cars and robots to power networks and bacterial metabolism. It studies dynamical systems—systems whose properties evolve over time—and focuses on how to analyze and control their behavior to achieve desirable outcomes, such as preventing car crashes, maintaining voltage levels, or positioning robots accurately. In this talk, we will explore at a high level how control theory can intersect with the study of natural language. First, we will discuss “Language To Control,” which explores how to integrate established control strategies with language algorithms. The goal is to enable intuitive communication with machines using natural language while ensuring the safety and reliability provided by classical control techniques. Second, we will delve into “Control For Language,” where we treat language production as a dynamical system and apply control theory to enhance our understanding and design of language technologies. This includes both foundational models and post-training methods. The aim of this talk is to demonstrate the potential of control theory as a tool for studying language and to open a discussion about potential future avenues.