Skip to main content
Browse by:

Machine Learning Seminar: How Many Directions Determine a Shape: a VC theory for modulii spaces of shapes

Sayan EC Slide
Wednesday, September 26, 2018
3:30 pm - 4:30 pm
Sayan Mukherjee
Machine Learning Seminar

We study transformations of shapes into representations that allow for analysis using standard statistical tools. The transformations are based on Euler integration and are of interest for their mathematical properties as well as their applications to science and engineering, because they provide a way of summarizing shapes in a topological, yet quantitative, way. By using an inversion theorem, we show that both transforms are injective on the space of shapes---each shape has a unique transform. By making use of a stratified space structure on the sphere, induced by hyperplane divisions, we prove additional uniqueness results in terms of distributions on the space of Euler curves. The main theoretical result provides the first (to our
knowledge) finite bound required to specify any shape in certain uncountable families of shapes, bounded below by curvature. This result is perhaps best appreciated in terms of shattering number or the perspective that any point in these particular moduli spaces of shapes is indexed using a tree of finite depth. We also show how these transformations can be used in practice for medical imaging applications as well as for evolutionary morphology questions.

Contact: Ariel Dawn