Learning Structural Node Representations Using Graph Kernels

Many applications require identifying nodes that perform similar functions in a graph. For instance, identifying structurally equivalent nodes can provide insight into the structure of complex networks. Learning latent representations that capture such structural role information about nodes has recently gained a lot of attention. Existing techniques for learning such representations typically rely on manually engineered features or are very expensive in terms of time and memory requirements. In this paper, we propose SEGK, a powerful framework for computing structural node representations. SEGK learns node representations by generating (or approximating) and decomposing a kernel matrix that incorporates structural similarity between nodes. To compute the similarity between two nodes, the proposed framework builds on well-established concepts from graph mining. Specifically, it compares the neighborhood subgraphs of increasing size of two nodes using graph kernels. SEGK is very flexible, and besides unlabeled graphs, it can also handle node-labeled and node-attributed graphs. We evaluate the proposed framework on several synthetic and real-world datasets, and compare its performance to state-of-the-art techniques for learning structural node embeddings. In almost all cases, the instances of the proposed framework outperform the competing methods, while their time complexity remains very attractive.

[1]  Kurt Mehlhorn,et al.  Efficient graphlet kernels for large graph comparison , 2009, AISTATS.

[2]  Danai Koutra,et al.  RolX: structural role extraction & mining in large graphs , 2012, KDD.

[3]  Ryan A. Rossi,et al.  Role Discovery in Networks , 2014, IEEE Transactions on Knowledge and Data Engineering.

[4]  R. Merton Social Theory and Social Structure , 1958 .

[5]  Jure Leskovec,et al.  Learning Structural Node Embeddings via Diffusion Wavelets , 2017, KDD.

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

[7]  Massimo Bartoletti,et al.  Data Mining for Detecting Bitcoin Ponzi Schemes , 2018, 2018 Crypto Valley Conference on Blockchain Technology (CVCBT).

[8]  Ian Davidson,et al.  Guided learning for role discovery (GLRD): framework, algorithms, and applications , 2013, KDD.

[9]  Emmanuel Müller,et al.  VERSE: Versatile Graph Embeddings from Similarity Measures , 2018, WWW.

[10]  Michalis Vazirgiannis,et al.  Matching Node Embeddings for Graph Similarity , 2017, AAAI.

[11]  Philip S. Yu,et al.  Deep Recursive Network Embedding with Regular Equivalence , 2018, KDD.

[12]  Steven Skiena,et al.  DeepWalk: online learning of social representations , 2014, KDD.

[13]  Daniel R. Figueiredo,et al.  struc2vec: Learning Node Representations from Structural Identity , 2017, KDD.

[14]  Jian Pei,et al.  Community Preserving Network Embedding , 2017, AAAI.

[15]  Qiongkai Xu,et al.  GraRep: Learning Graph Representations with Global Structural Information , 2015, CIKM.

[16]  Michalis Vazirgiannis,et al.  Graph Kernels: A Survey , 2019, J. Artif. Intell. Res..

[17]  Jie Wu,et al.  Small Worlds: The Dynamics of Networks between Order and Randomness , 2003 .

[18]  Mark Heimann,et al.  REGAL: Representation Learning-based Graph Alignment , 2018, CIKM.

[19]  Hans-Peter Kriegel,et al.  Shortest-path kernels on graphs , 2005, Fifth IEEE International Conference on Data Mining (ICDM'05).

[20]  Kristian Kersting,et al.  Faster Kernels for Graphs with Continuous Attributes via Hashing , 2016, 2016 IEEE 16th International Conference on Data Mining (ICDM).

[21]  Kevin Chen-Chuan Chang,et al.  A Comprehensive Survey of Graph Embedding: Problems, Techniques, and Applications , 2017, IEEE Transactions on Knowledge and Data Engineering.

[22]  Jure Leskovec,et al.  node2vec: Scalable Feature Learning for Networks , 2016, KDD.

[23]  Dragomir R. Radev,et al.  The ACL anthology network corpus , 2009, Language Resources and Evaluation.

[24]  Cody Buntain,et al.  Identifying social roles in reddit using network structure , 2014, WWW.

[25]  Yang Liu,et al.  subgraph2vec: Learning Distributed Representations of Rooted Sub-graphs from Large Graphs , 2016, ArXiv.