Learning Spectral Graph Transformations for Link Prediction

Abstract

We present a unified framework for learning link prediction and edge weight prediction functions in large networks, based on the transformation of a graph's algebraic spectrum. Our approach generalizes several graph kernels and dimensionality reduction methods and provides a method to estimate their parameters efficiently. We show how the parameters of these prediction functions can be learned by reducing the problem to a one-dimensional regression problem whose runtime only depends on the method's reduced rank and that can be inspected visually. We derive variants that apply to undirected, weighted, unweighted, unipartite and bipartite graphs. We evaluate our method experimentally using examples from social networks, collaborative filtering, trust networks, citation networks, authorship graphs and hyperlink networks.

@inproceedings{1553447,
 author = {Kunegis, J'{e}r^{o}me and Lommatzsch, Andreas},
 title = {Learning spectral graph transformations for link prediction},
 booktitle = {ICML '09: Proceedings of the 26th Annual International Conference on Machine Learning},
 year = {2009},
 isbn = {978-1-60558-516-1},
 pages = {1--8},
 location = {Montreal, Quebec, Canada},
 doi = {http://doi.acm.org/10.1145/1553374.1553447},
 publisher = {ACM},
 address = {New York, NY, USA},
 }
Autoren:
Jérôme Kunegis, Andreas Lommatzsch
Kategorie:
Tagungsbeitrag
Jahr:
2009
Ort:
26th International Conference on Machine Learning