Hypergraph Random Walks, Laplacians, and Clustering

Conference · Mon Oct 19 04:00:00 EDT 2020

DOI:https://doi.org/10.1145/3340531.3412034· OSTI ID:1691481

Hayashi, Koby ^[1]; Aksoy, Sinan G. ^[2]; Park, Cheong Hee ^[3]; Park, Haesun ^[4]

---

1. Georgia Institute of Technology
2. BATTELLE (PACIFIC NW LAB)
3. Chungnam National University
4. Academia

We propose a flexible framework for clustering hypergraph-structured data based on recently proposed random walks utilizing edge-dependent vertex weights. When incorporating edge-dependent vertex weights (EDVW), a weight is associated with each vertex-hyperedge pair, yielding a weighted incidence matrix of the hypergraph. Such weightings have been utilized in term-document representations of text data sets. We explain how random walks with EDVW serve to construct different hypergraph Laplacian matrices, and then develop a suite of clustering methods that use these incidence matrices and Laplacians for hypergraph clustering. Using 20Newsgroup, U.S. patent, Reuters' Corpus Volume 1, and genetics data sets, we compare the performance of these clustering algorithms experimentally against a variety of existing hypergraph clustering methods. We show that the proposed methods produce higher-quality clusters.

• Research Organization: Pacific Northwest National Laboratory (PNNL), Richland, WA (United States)
• Sponsoring Organization: USDOE
• DOE Contract Number: AC05-76RL01830
• OSTI ID: 1691481
• Report Number(s): PNNL-SA-153213
• Country of Publication: United States
• Language: English

Related Subjects

hypergraphs
clustering
laplacian
random walk
Markov Chain