论文信息 - Measuring the Expressivity of Graph Kernels through the Rademacher Complexity

Measuring the Expressivity of Graph Kernels through the Rademacher Complexity

Graph kernels are widely adopted in real-world applications that involve learning on graph data. Different graph kernels have been proposed in literature, but no theoretical comparison among them is present. In this paper we provide a formal definition for the expressiveness of a graph kernel by means of the Rademacher Complexity, and analyze the differences among some state-of-the-art graph kernels. Results on real world datasets confirm some known properties of graph kernels, showing that the Rademacher Complexity is indeed a suitable measure for this analysis.

[1] Nello Cristianini,et al. Kernel Methods for Pattern Analysis , 2003, ICTAI.

[2] Alessandro Sperduti,et al. Exploiting the ODD framework to define a novel effective graph kernel , 2015, ESANN.

[3] Vladimir Vapnik,et al. Statistical learning theory , 1998 .

[4] M. Talagrand. The Glivenko-Cantelli Problem , 1987 .

[5] Alessandro Sperduti,et al. A Tree-Based Kernel for Graphs , 2012, SDM.

[6] Peter L. Bartlett,et al. Rademacher and Gaussian Complexities: Risk Bounds and Structural Results , 2003, J. Mach. Learn. Res..

[7] Davide Anguita,et al. Tikhonov, Ivanov and Morozov regularization for support vector machine learning , 2015, Machine Learning.

[8] Kurt Mehlhorn,et al. Weisfeiler-Lehman Graph Kernels , 2011, J. Mach. Learn. Res..

[9] Thomas Gärtner,et al. On Graph Kernels: Hardness Results and Efficient Alternatives , 2003, COLT.

[10] Davide Anguita,et al. Local Rademacher Complexity: Sharper risk bounds with and without unlabeled samples , 2015, Neural Networks.

[11] Alessandro Sperduti,et al. A memory efficient graph kernel , 2012, The 2012 International Joint Conference on Neural Networks (IJCNN).

[12] Fabrizio Costa,et al. Fast Neighborhood Subgraph Pairwise Distance Kernel , 2010, ICML.

[13] Davide Anguita,et al. Global Rademacher Complexity Bounds: From Slow to Fast Convergence Rates , 2015, Neural Processing Letters.

[14] P. Bartlett,et al. Local Rademacher complexities , 2005, math/0508275.