Detecting Equivalence Between Iterative Algorithms for Optimization
Abstract
The talk presents a framework for reasoning about equivalence between a broad class of iterative algorithms, with a focus on algorithms designed for convex optimization. The framework defines several notions of algorithm equivalence and provides computationally tractable means to detect equivalence. The speaker introduces the concept of oracle equivalence, which states that two algorithms are equivalent if they result in the same sequence of calls to the function oracles. The framework utilizes state-space realizations to represent algorithms and characterizes algorithm equivalence via transfer functions. The speaker also discusses the software package "Linnaeus" that implements the framework and makes it easy to find equivalent iterative algorithms.
Bio
Madeleine Udell is an Assistant Professor of Operations Research and Information Engineering at Cornell University. Her research focuses on optimization and machine learning for large-scale data analysis and control, with applications in various fields. She has received several awards for her work, including the Alfred P. Sloan Research Fellowship, the National Science Foundation CAREER award, and the Office of Naval Research Young Investigator Award.
