Robust Graph Mode Seeking by Graph Shift
Abstract
In this paper, we study how to robustly computethe modes of a graph, namely the densesubgraphs, which characterize the underlyingcompact patterns and are thus useful formany applications. We first define the modesbased on graph density function, then proposethe graph shift algorithm, which startsfrom each vertex and iteratively shifts towardsthe nearest mode of the graph alonga certain trajectory. Both theoretic analysisand experiments show that graph shift algorithmis very efficient and robust, especiallywhen there exist large amount of noises andoutliers.
Cite
Text
Liu and Yan. "Robust Graph Mode Seeking by Graph Shift." International Conference on Machine Learning, 2010.Markdown
[Liu and Yan. "Robust Graph Mode Seeking by Graph Shift." International Conference on Machine Learning, 2010.](https://mlanthology.org/icml/2010/liu2010icml-robust-a/)BibTeX
@inproceedings{liu2010icml-robust-a,
title = {{Robust Graph Mode Seeking by Graph Shift}},
author = {Liu, Hairong and Yan, Shuicheng},
booktitle = {International Conference on Machine Learning},
year = {2010},
pages = {671-678},
url = {https://mlanthology.org/icml/2010/liu2010icml-robust-a/}
}