Scalable kernels for graphs with continuous attributes

By Aasa Feragen, Niklas Kasenburg, Jens Petersen, Marleen de Bruijne, Karsten M. Borgwardt. 2013

191 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Hierarchical Graph Representation Learning with Differentiable Pooling (2018, methodology): “For ENZYMES, COLLAB, and PROTEINS data sets, the number of hidden dimensions is set to 64.”
Covariant Compositional Networks For Learning Graphs (2018, neutral): “…on graphs has a long history in the kernels literature, including approaches based on random walks (G ¨ artner, 2002; Borgwardt & Kriegel, 2005; Feragen et al., 2013), counting subgraphs (Shervashidze et al., 2009), spectral ideas (Vishwanathan et al., 2010), label propagation schemes with…”
Propagation kernels: efficient graph kernels from propagated information (2014, background): “Kernels for graphs with continuous node attributes have only recently gained greater interest (Borgwardt and Kriegel 2005; Kriege and Mutzel 2012; Feragen et al. 2013).”
The Multiscale Laplacian Graph Kernel (2016, background): “…different graph kernels are motivated in very different ways from random walks (G¨artner, 2002) through shortest paths (Borgwardt & Kriegel, 2005; Feragen et al., 2013) to Fourier transforms on the symmetric group (Kondor & Borgwardt, 2008), ultimately most graph kernels also reduce to computing…”
Fast Memory-efficient Anomaly Detection in Streaming Heterogeneous Graphs (2016, background): “Graph kernels [26, 33, 13] decompose each graph into a set of local substructures and the similarity between two graphs is a function of the number of substructures they have in common.”
Glocalized Weisfeiler-Lehman Graph Kernels: Global-Local Feature Maps of Graphs (2017, methodology): “Note that this is the same data set as used in [52], which does not contain all the annotations described and used in [51].”
A survey on graph kernels (2019, methodology): “Another graph kernel, which fits into the framework of weighted vertex kernels, is the GraphHopper kernel (Feragen et al. 2013) with Here M ( v ) and M ( v (cid:7) ) are δ × δ matrices, where the entry M ( v ) ij for v in V ( G ) counts the number of times the vertex v appears as the i -th vertex on”
Deep Graph Kernels (2015, background): “One can show that all sub-paths of a shortest-path are also shortest-paths with the same source [8].”
Graph Capsule Convolutional Neural Networks (2018, background): “Another line of research in this area focuses on efﬁciently computing these kernels either through exploiting certain structure dependency, or via approximation or randomization (Feragen et al., 2013; de Vries, 2013; Neumann et al., 2012).”
Graph Invariant Kernels (2015, methodology): “Graphs with continuous attributes have been much less investigated, but one graph kernel that handles continuous attributes, is based on random walks (RW) labels [Kashima et al., 2003].”
An Aligned Subtree Kernel for Weighted Graphs (2015, background): “In general, note that any graph kernel what can deal with continuous node attributes (Neumann et al., 2012; Feragen et al., 2013) can be adapted to work on edge weighted graphs by replacing the original graph with its corresponding line graph, effectively moving the weights from the edges to the nod”
RetGK: Graph Kernels based on Return Probabilities of Random Walks (2018, background): “SYNTHETIC and Synthie are synthetic datasets based on random graphs, which were first introduced in [9] and [19], respectively.”
Matching Node Embeddings for Graph Similarity (2017, methodology): “Graph kernels have been successfully used for comparing graphs in a variety of fields such as Computational Biology (Schölkopf, Tsuda, and Vert 2004), Chemistry (Mahé and Vert 2009), Information Retrieval (Hermansson et al. 2013) and Medicine (Feragen et al. 2013).”
Tree-Based Kernel for Graphs With Continuous Attributes (2015, methodology): “presented in Section V-C, the Propagation kernel ( P2K) [11], different instantiations of the Graph Invariant Kernels [1 2] (GIK NSPDK, GIK SGK GLOBAL, GIK SGK LOCAL, GIK WL GLOBAL, GIK WL LOCAL) and the results from the paper [10], corrected according to the erratum.”
InfoGraph: Unsupervised and Semi-supervised Graph-Level Representation Learning via Mutual Information Maximization (2019, background): “Another line of research in this area focuses on efﬁciently computing these kernels either through exploiting certain structural dependencies, or via approximations/randomization [11, 9, 37].”
TUDataset: A collection of benchmark datasets for learning with graphs (2020, methodology): “The datasets S YNTHETIC N EW and S YNTHIE were created by Feragen et al. (2013) (see Erratum) and Morris et al. (2016), respectively, to demonstrate the ability of kernels to operate on graphs with continuous attributes.”
Optimal Transport for structured data with application on graphs (2018, methodology): “For real valued vector attributes, we consider the HOPPER kernel (HOPPERK) (Feragen et al., 2013) and the propagation kernel (PROPAK) (Neumann et al.”