ABSTRACT
The central issue in representing graph-structured data instances in learning algorithms is designing features which are invariant to permuting the numbering of the vertices. We present a new system of invariant graph features which we call the skew spectrum of graphs. The skew spectrum is based on mapping the adjacency matrix of any (weigted, directed, unlabeled) graph to a function on the symmetric group and computing bispectral invariants. The reduced form of the skew spectrum is computable in O(n3) time, and experiments show that on several benchmark datasets it can outperform state of the art graph kernels.
- Bonchev, D., & Rouvray, D. H. (Eds.). (1991). Chemical graph theory: Introduction and fundamentals, vol. 1. London, UK: Gordon and Breach Science Publishers.Google Scholar
- Borgwardt, K. M., & Kriegel, H.-P. (2005). Shortest-path kernels on graphs. Proc. Intl. Conf. Data Mining (pp. 74--81). Google ScholarDigital Library
- Borgwardt, K. M., Ong, C. S., Schonauer, S., Vishwanathan, S. V. N., Smola, A. J., & Kriegel, H. P. (2005). Protein function prediction via graph kernels. Bioinformatics, 21, i47--i56. Google ScholarDigital Library
- Chang, C.-C., & Lin, C.-J. (2001). LIBSVM: a library for support vector machines. Software available at http://www.csie.ntu.edu.tw/~cjlin/libsvm. Google ScholarDigital Library
- Clausen, M. (1989). Fast generalized Fourier transforms. Theor. Comput. Sci., 55--63. Google ScholarDigital Library
- Collins, M., & Duffy, N. (2002). Convolution kernels for natural language. Advances in Neural Information Processing Systems 14. Cambridge, MA: MIT Press.Google Scholar
- Debnath, A. K., Lopez de Compadre, R. L., Debnath, G., Shusterman, A. J., & Hansch, C. (1991). Structure-activity relationship of mutagenic aromatic and heteroaromatic nitro compounds. correlation with molecular orbital energies and hydrophobicity. J Med Chem, 34, 786--797.Google ScholarCross Ref
- Diaconis, P. (1988). Group representation in probability and statistics, vol. 11 of IMS Lecture Series. Institute of Mathematical Statistics.Google Scholar
- Gärtner, T. (2003). A survey of kernels for structured data. SIGKDD Explorations, 5, 49--58. Google ScholarDigital Library
- Gärtner, T., Flach, P., & Wrobel, S. (2003). On graph kernels: Hardness results and efficient alternatives. COLT03 (pp. 129--143). Springer.Google Scholar
- Harchaoui, Z., & Bach, F. (2007). Image classification with segmentation graph kernels. Proceedings of CVPR07.Google ScholarCross Ref
- Huang, J., Guestrin, C., & Guibas, L. (2008). Efficient inference for distributions on permutations. Proceedings of NIPS07.Google Scholar
- James, G., & Kerber, A. (1981). The representation theory of the symmetric group. Addison-Wesley.Google Scholar
- Kakarala, R. (1992). Triple corelation on groups. Doctoral dissertation, Department of Mathematics, UC Irvine. Google ScholarDigital Library
- Kakarala, R. (1993). A group theoretic approach to the triple correlation. IEEE Workshop on higher order statistics (pp. 28--32).Google ScholarCross Ref
- Kondor, R. (2007a). A novel set of rotationally and translationally invariant features for images based on the non-commutative bispectrum. http://arxiv.org/abs/cs.CV/0701127.Google Scholar
- Kondor, R. (2007b). The skew spectrum of functions on finite groups and their homogeneous spaces. http://arxiv.org/abs/0712.4259.Google Scholar
- Kondor, R. (2008). The skew spectrum of graphs. To appear at http://arxiv.org/.Google Scholar
- Kondor, R., Howard, A., & Jebara, T. (2007). Multi-object tracking with representations of the symmetric group. Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics.Google Scholar
- Kumar, R., Novak, J., & Tomkins, A. (2006). Structure and evolution of online social networks. KDD (pp. 611--617). Google ScholarDigital Library
- Rockmore, D. N. (1997). Some applications of generalized FFTs. Proceedings of the DIMACS workshop on groups and computation.Google ScholarCross Ref
- Schomburg, I., Chang, A., Ebeling, C., Gremse, M., Heldt, C., Huhn, G., & Schomburg, D. (2004). Brenda, the enzyme database: updates and major new developments. Nucleic Acids Research, 32D, 431--433.Google ScholarCross Ref
- Sharan, R., & Ideker, T. (2006). Modeling cellular machinery through biological network comparison. Nature Biotechnology, 24, 427--433.Google ScholarCross Ref
- Shawe-Taylor, J. (1993). Symmetries and discriminability in feedforward network architectures. IEEE Transactions on Neural Networks, 4, 816--826.Google ScholarDigital Library
- Wale, N., & Karypis, G. (2006). Comparison of descriptor spaces for chemical compound retrieval and classification. Proc. of ICDM (pp. 678--689). Hong Kong. Google ScholarDigital Library
Index Terms
- The skew spectrum of graphs
Recommendations
Perfect Graphs with No Balanced Skew-Partition are 2-Clique-Colorable
A graph G is perfect if for all induced subgraphs H of G, ï H=ωH. A graph G is Berge if neither G nor its complement contains an induced odd cycle of length at least five. The Strong Perfect Graph Theoremï [9] states that a graph is perfect if and only ...
Bicyclic oriented graphs with skew-rank 6
Let G be an oriented graph and S(G ) be its skew-adjacency matrix. The skew-rank of G , denoted by sr(G ), is the rank of S(G ). In this paper, we characterize all the bicyclic oriented graphs with skew-rank 6. Let G be a bicyclic oriented graph with ...
Clique-Stable Set separation in perfect graphs with no balanced skew-partitions
Inspired by a question of Yannakakis on the Vertex Packing polytope of perfect graphs, we study the Clique-Stable Set separation in a non-hereditary subclass of perfect graphs. A cut ( B , W ) of G (a bipartition of V ( G ) ) separates a clique K and a ...
Comments