Classification of Euler Characteristic Transforms Using CNNs - 5/9/24

Explore the application of convolutional neural networks (CNNs) to classify Euler Characteristic Transforms (ECT) in this insightful 48-minute talk by Sarah McGuire from the Applied Algebraic Topology Network. Delve into the advantages of ECT as a descriptive method for representing topological shape data, comparing it to the Persistent Homology transform. Learn about a proposed variant CNN architecture specifically designed for ECT data classification, taking advantage of its cylindrical structure. Discover the important rotation equivariance properties of this model and examine its practical applications through two leaf-shape datasets. Gain valuable insights into this innovative approach that combines topological data analysis with deep learning techniques for shape classification tasks.

Sarah McGuire (5/9/24): Classification of Euler Characteristic Transforms using CNNs