Principal persistence measures compared by MMD yield a scalable, GPU-parallel topological regularizer with claimed continuous gradients, improving GAN image generation and semi-supervised classification.
Persistence by Parts: Multiscale Feature Detection via Distributed Persistent Homology
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
A method is presented for the distributed computation of persistent homology, based on an extension of the generalized Mayer-Vietoris principle to filtered spaces. Cellular cosheaves and spectral sequences are used to compute global persistent homology based on local computations indexed by a scalar field. These techniques permit computation localized not merely by geography, but by other features of data points, such as density. As an example of the latter, the construction is used in the multi-scale analysis of point clouds to detect features of varying sizes that are overlooked by standard persistent homology.
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Towards Scalable Topological Regularizers
Principal persistence measures compared by MMD yield a scalable, GPU-parallel topological regularizer with claimed continuous gradients, improving GAN image generation and semi-supervised classification.