Machine Learning Meets Geometry

CSE291-F00 - Winter 2020

Schedule and assignments

Unit 1: Theories of Geometry

1/7	Numerical Methods (I)	overview of the course, logistics
1/9	Numerical Methods (II)	linear system, optimization.
1/14	Curves (I)	curve theory, Frenet frame. Reference: Intro to DG, Ch2, Board Derivations
1/16	Curves (II)	Gauss Map, Turning Number Theorem, Bishop Frame. Reference: Intro to DG, Ch2
1/21	Shape Representation Basics	point cloud, parametric surface, mesh, implicit surface, point cloud to implicit functions
1/23	Course Project Introductions	introduced course project policies and possible projects
1/28	Shape Representations (II)	point cloud, parametric surface, mesh, implicit surface, point cloud to implicit functions
1/30	Surface Curvature	second fundamental form, gaussian curvature. slides credit: Mira Ben-Chen's course in CS468 at Stanford and 6.8383 by Justin Solomon at MIT. References: Intro to DG, Ch3, 4, 5
2/4	Computation of Surface Curvature	second fundamental form, gaussian curvature. slides credit: Keenan Crane at CMU and Justin Solomon at MIT. References: Intro to DG, Ch3, 4, 5
2/6	Point Cloud Reconstruction	Earth Mover's Distance, point cloud reconstruction
2/11	Point Cloud Reconstruction II	cont. of point cloud reconstruction, deformation based approach
2/13	Geodesic Distances	continuous theory of geodesics, fast marching algorithm
2/18	Laplacian Smoothing	motivating applications of Laplacian: smoothing, cotangent laplacian
2/20	Laplacian Mesh Editing and Spectral Graph Theory	motivating applications of Laplacian: mesh editing, basics of spectral graph theory
3/3	Intrinsic Shape Feature	data embedding, intrinsic shape feature, heat kernel signature
3/5	Deep Learning for 3D Recognition	PointNet, SparseConv, VoteNet
3/10	Continuous Laplacian, Functional Map, Spectral CNN	Continuous Laplacian, Functional Map, Spectral CNN