Scaffoldings and spines: organizing high-dimensional data using cover trees, local principal component analysis, and persistent homology
From MaRDI portal
Abstract: We propose a flexible and multi-scale method for organizing, visualizing, and understanding datasets sampled from or near stratified spaces. The first part of the algorithm produces a cover tree using adaptive thresholds based on a combination of multi-scale local principal component analysis and topological data analysis. The resulting cover tree nodes consist of points within or near the same stratum of the stratified space. They are then connected to form a emph{scaffolding} graph, which is then simplified and collapsed down into a emph{spine} graph. From this latter graph the stratified structure becomes apparent. We demonstrate our technique on several synthetic point cloud examples and we use it to understand song structure in musical audio data.
Recommendations
- A higher-dimensional homologically persistent skeleton
- Local homology transfer and stratification learning
- Topological pattern recognition for point cloud data
- Geometric diffusions as a tool for harmonic analysis and structure definition of data: diffusion maps
- Approximating local homology from samples
Cited in
(7)- The number of simply-connected trivalent 2-dimensional stratifolds
- 2-stratifold spines of closed 3-manifolds
- Models of simply-connected trivalent 2-dimensional stratifolds
- 2-stratifolds with fundamental group \(\mathbb{Z} \)
- String representation of trivalent 2-stratifolds with trivial fundamental group
- On the homeomorphism problem of trivalent 2-stratifolds with finite homology groups
- Topological fingerprints for audio identification
This page was built for publication: Scaffoldings and spines: organizing high-dimensional data using cover trees, local principal component analysis, and persistent homology
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1713432)
