Geometry, Topology and Discrete Symmetries Revealed by Deep Neural Networks
Offered By: Institut des Hautes Etudes Scientifiques (IHES) via YouTube
Course Description
Overview
Explore the intersection of universality efforts and manifold learning in this 36-minute lecture by Maarten de Hoop from Rice University. Delve into the question of which neural network architectures can universally approximate maps between topologically interesting manifolds. Examine the low-dimensional manifold hypothesis underlying inverse problems and its implications for data analysis. Discover a novel network architecture combining injective flows and coordinate projections that universally approximates local diffeomorphisms between compact smooth submanifolds in Euclidean spaces. Learn how this network enables multi-valued inversion and handles cases where the target map changes topology. Investigate applications in supervised learning for recovering group actions of finite group invariant maps and in unsupervised learning for informing topologically expressive starting spaces in generative models.
Syllabus
Maarten de Hoop - Geometry, topology and discrete symmetries revealed by deep neural networks
Taught by
Institut des Hautes Etudes Scientifiques (IHES)
Related Courses
An Introduction to Functional AnalysisÉcole Centrale Paris via Coursera Nonlinear Dynamics 1: Geometry of Chaos
Georgia Institute of Technology via Independent Topology in Condensed Matter: Tying Quantum Knots
Delft University of Technology via edX Математика для всех
Moscow Institute of Physics and Technology via Coursera Геометрия и группы
Moscow Institute of Physics and Technology via Coursera