معرفی
Kathryn Lindsey is Associate Professor of Mathematics at Boston College, researching neural network theory and dynamical systems. Her NSF-funded work investigates geometric foundations of deep learning and topological entropy in dynamical systems. She earned her Ph.D. from Cornell University and teaches courses in analysis and geometry.
Lindsey's mathematical research has two primary strands: analyzing ReLU neural networks as piecewise linear function spaces and studying Thurston sets in complex dynamics. She develops topological measures for network expressivity and entropy algorithms for Mandelbrot sets. Recent publications feature functional dimension of neural architectures and symmetry detection in parameter spaces.
Publications demonstrate dual focus: neural network theory articles examine functional dimension, transversality, and symmetry in ReLU networks, while dynamics papers characterize Master Teapot geometry and bicritical iterates. Methodologies blend combinatorics, topology, and complex analysis.
- Faculty Teaching Award, National University of Singapore (2007)
- NSF Grant #2133822 for neural network research
Lindsey actively advises doctoral students (Laura Seaberg, Yaoying Fu) and has supervised seven thesis students. She directs the graduate research seminar and maintains collaborations through conferences like CIMAT's Applied Topology School.
