Source author record

Jan Reininghaus

Jan Reininghaus appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

math.AT Computational Geometry Computer Vision Discrete Mathematics Distributed, Parallel, and Cluster Computing Machine Learning physics.flu-dyn

Catalog footprint

What is connected

5works

7topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2014arXiv

A Stable Multi-Scale Kernel for Topological Machine Learning

Topological data analysis offers a rich source of valuable information to study vision problems. Yet, so far we lack a theoretically sound connection to popular kernel-based learning techniques, such as kernel SVMs or kernel PCA. In this work, we establish such a connection by designing a multi-scale kernel for persistence diagrams, a stable summary representation of topological features in data. We show that this kernel is positive definite and prove its stability with respect to the 1-Wasserstein distance. Experiments on two benchmark datasets for 3D shape classification/retrieval and texture recognition show considerable performance gains of the proposed method compared to an alternative approach that is based on the recently introduced persistence landscapes.

preprint2014arXiv

Acceleration feature points of unsteady shear flows

In this paper, we propose a novel framework to extract features such as vortex cores and saddle points in two-dimensional unsteady flows. This feature extraction strategy generalizes critical points of snapshot topology in a Galilean-invariant manner, allows to prioritize features according to their strength and longevity, enables to track the temporal evolution of features, is robust against noise and has no subjective parameters. These characteristics are realized via several constitutive elements. First, acceleration is employed as a feature identifier following Goto and Vassilicos (2006), thus ensuring Galilean invariance. Second, the acceleration magnitude is used as basis for a mathematically well-developed scalar field topology. The minima of this field are called acceleration feature points, a superset of the acceleration zeros. These points are discriminated into vortices and saddle points depending the spectral properties of the velocity Jacobian. Third, all operations are based on discrete topology for the scalar field with combinatorial algorithms. This parameter-free foundation allows (1) to use persistence as a physically meaningful importance measure to prioritize feature points, (2) ensures robustness since no differentiation and interpolation need to be performed with the data, and (3) enables a natural and robust tracking algorithm for the temporal feature evolution. In particular, we can track vortex merging events in an unsupervised manner. Data based analyses are presented for an incompressible periodic cylinder wake, an incompressible planar mixing layer and a weakly compressible planar jet. They demonstrate the power of the tracking approach, which provides a spatiotemporal hierarchy of the minima.

preprint2013arXiv

Clear and Compress: Computing Persistent Homology in Chunks

We present a parallelizable algorithm for computing the persistent homology of a filtered chain complex. Our approach differs from the commonly used reduction algorithm by first computing persistence pairs within local chunks, then simplifying the unpaired columns, and finally applying standard reduction on the simplified matrix. The approach generalizes a technique by Günther et al., which uses discrete Morse Theory to compute persistence; we derive the same worst-case complexity bound in a more general context. The algorithm employs several practical optimization techniques which are of independent interest. Our sequential implementation of the algorithm is competitive with state-of-the-art methods, and we improve the performance through parallelized computation.

preprint2013arXiv

Distributed computation of persistent homology

Persistent homology is a popular and powerful tool for capturing topological features of data. Advances in algorithms for computing persistent homology have reduced the computation time drastically -- as long as the algorithm does not exhaust the available memory. Following up on a recently presented parallel method for persistence computation on shared memory systems, we demonstrate that a simple adaption of the standard reduction algorithm leads to a variant for distributed systems. Our algorithmic design ensures that the data is distributed over the nodes without redundancy; this permits the computation of much larger instances than on a single machine. Moreover, we observe that the parallelism at least compensates for the overhead caused by communication between nodes, and often even speeds up the computation compared to sequential and even parallel shared memory algorithms. In our experiments, we were able to compute the persistent homology of filtrations with more than a billion (10^9) elements within seconds on a cluster with 32 nodes using less than 10GB of memory per node.

preprint2012arXiv

Combinatorial Gradient Fields for 2D Images with Empirically Convergent Separatrices

This paper proposes an efficient probabilistic method that computes combinatorial gradient fields for two dimensional image data. In contrast to existing algorithms, this approach yields a geometric Morse-Smale complex that converges almost surely to its continuous counterpart when the image resolution is increased. This approach is motivated using basic ideas from probability theory and builds upon an algorithm from discrete Morse theory with a strong mathematical foundation. While a formal proof is only hinted at, we do provide a thorough numerical evaluation of our method and compare it to established algorithms.