Traditional machine learning, pattern recognition and data analysis methods often assume that input data can be represented well by elements of Euclidean space. While this assumption has worked well for many past applications, researchers have increasingly realized that most data in vision and pattern recognition is intrinsically non-Euclidean, i.e. standard Euclidean calculus does not apply. The exploitation of this geometrical information can lead to more accurate representation of the inherent structure of the data, better algorithms, and better performance in practical applications. In particular, Riemannian geometric principles can be applied to a variety of difficult computer vision problems including face recognition, activity recognition, object detection, biomedical image analysis, and structure-from-motion, to name a few. Consequently, Riemannian geometric computing has become increasingly popular in the computer vision community. Besides nice mathematical formulations, Riemannian computations based on the geometry of underlying manifolds are often faster and more stable than their classical counterparts. Over the past few years, the popularity of Riemannian algorithms has increased several-fold. Some of the mathematical entities that benefit from a geometric analysis include rotation matrices, medial representations, subspace comparisons, symmetric positive-definite matrices, function spaces, and more. This workshop aims to bring together the best emerging work in this interdisciplinary field, with topics of wide interest to computer vision and machine learning.
Main topics of interest
- Deep learning and geometry.
- Riemannian methods in computer vision
- Statistical shape analysis: detection, estimation, and inference.
- Statistical analysis on manifolds
- Manifold-valued features and learning
- Machine learning on nonlinear manifolds
- Shape detection, tracking and retrieval.
- Topological methods and structure analysis
- Functional Data Analysis: Hilbert manifolds, Visualization.
- Applications: Medical analysis, Biometrics, Biology, Environmetrics, Graphics, Activity recognition, Bioinformatics, Pattern recognition, etc.