A jensen-shannon kernel for hypergraphs

Authors:
Lu Bai;Edwin R. Hancock;Peng Ren
Affiliations:
Department of Computer Science, University of York, UK, Heslington, York, UK;Department of Computer Science, University of York, UK, Heslington, York, UK;College of Information and Control Engineering, China University of Petroleum (Huadong), China
Venue:
SSPR'12/SPR'12 Proceedings of the 2012 Joint IAPR international conference on Structural, Syntactic, and Statistical Pattern Recognition
Year:
2012

Citing 8
Cited 0

Modeling the Shape of the Scene: A Holistic Representation of the Spatial Envelope

International Journal of Computer Vision
Shortest-Path Kernels on Graphs

ICDM '05 Proceedings of the Fifth IEEE International Conference on Data Mining
Learning from interpretations: a rooted kernel for ordered hypergraphs

Proceedings of the 24th international conference on Machine learning
Nonextensive Information Theoretic Kernels on Measures

The Journal of Machine Learning Research
Efficient High Order Matching

IEEE Transactions on Pattern Analysis and Machine Intelligence
A polynomial characterization of hypergraphs using the Ihara zeta function

Pattern Recognition
The cover times of random walks on hypergraphs

SIROCCO'11 Proceedings of the 18th international conference on Structural information and communication complexity
Weisfeiler-Lehman Graph Kernels

The Journal of Machine Learning Research

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper we explore how to construct a Jensen-Shannon kernel for hypergraphs. We commence by calculating probability distribution over the steady state random walk on a hypergraph. The Shannon entropies required to construct the Jensen-Shannon divergence for pairs of hypergraphs are obtained from steady state probability distributions of the random walk. The Jensen-Shannon divergence between a pair of hypergraphs is the difference between the Shannon entropies of the separate hypergraphs and a composite structure. Our proposed kernel is not restricted to hypergraphs. Experiments on (hyper)graph datasets extracted from bioinformatics and computer vision datasets demonstrate the effectiveness and efficiency of the Jensen-Shannon hypergraph kernel for classification and clustering.