ABSTRACT
The two predominant approaches to graph comparison in recent years are based on (i) enumerating matching subgraphs or (ii) comparing neighborhoods of nodes. In this work, we complement these two perspectives with a third way of representing graphs: using filtration curves from topological data analysis that capture both edge weight information and global graph structure. Filtration curves are highly efficient to compute and lead to expressive representations of graphs, which we demonstrate on graph classification benchmark datasets. Our work opens the door to a new form of graph representation in data mining.
Supplemental Material
- Kubilay Atasu and Thomas Mittelholzer. 2019. Linear-Complexity Data-Parallel Earth Mover's Distance Approximations. In ICML, Vol. 97. PMLR, 364--373.Google Scholar
- Serguei A. Barannikov. 1994. The Framed Morse Complex and its Invariants. Advances in Soviet Mathematics, Vol. 21 (1994), 93--115.Google Scholar
- Eric Berry, Yen-Chi Chen, Jessi Cisewski-Kehe, and Brittany Terese Fasy. 2018. Functional Summaries of Persistence Diagrams. arXiv e-prints, Article arXiv:1804.01618 (2018), arXiv:1804.01618 pages.arxiv: 1804.01618 [stat.ME]Google Scholar
- Karsten Borgwardt, Elisabetta Ghisu, Felipe Llinares-López, Leslie O'Bray, and Bastian Rieck. 2020. Graph Kernels: State-of-the-Art and Future Challenges. Foundations and Trends® in Machine Learning, Vol. 13, 5--6 (2020), 531--712.Google ScholarCross Ref
- Karsten Borgwardt and Hans-Peter Kriegel. 2005. Shortest-path kernels on graphs. In Proceedings of the Fifth IEEE International Conference on Data Mining. IEEE Computer Society, Washington, DC, USA, 74--81.Google ScholarDigital Library
- Karsten M. Borgwardt, Cheng Soon Ong, Stefan Schönauer, S. V. N. Vishwanathan, Alex J. Smola, and Hans-Peter Kriegel. 2005. Protein function prediction via graph kernels. Bioinformatics, Vol. 21, suppl 1 (2005), i47--i56.Google ScholarDigital Library
- Horst Bunke and Kaspar Riesen. 2007. A Family of Novel Graph Kernels for Structural Pattern Recognition. In Progress in Pattern Recognition, Image Analysis and Applications. Springer, Heidelberg, Germany, 20--31.Google Scholar
- Mathieu Carrière, Frédéric Chazal, Yuichi Ike, Théo Lacombe, Martin Royer, and Yuhei Umeda. 2020. PersLay: A Neural Network Layer for Persistence Diagrams and New Graph Topological Signatures. In Proceedings of the 23rd International Conference on Artificial Intelligence and Statistics, Vol. 108. PMLR, 2786--2796.Google Scholar
- Deli Chen, Yankai Lin, Wei Li, Peng Li, Jie Zhou, and Xu Sun. 2020. Measuring and Relieving the Over-Smoothing Problem for Graph Neural Networks from the Topological View. Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 34, 4 (2020), 3438--3445.Google ScholarCross Ref
- Ilya Chevyrev, Vidit Nanda, and Harald Oberhauser. 2020. Persistence Paths and Signature Features in Topological Data Analysis. IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 42, 1 (2020), 192--202.Google ScholarDigital Library
- Herbert Edelsbrunner, David Letscher, and Afra Zomorodian. 2002. Topological Persistence and Simplification. Discrete & Computational Geometry, Vol. 28, 4 (2002), 511--533.Google ScholarDigital Library
- Federico Errica, Marco Podda, Davide Bacciu, and Alessio Micheli. 2020. A Fair Comparison of Graph Neural Networks for Graph Classification. In ICLR .Google Scholar
- Justin Gilmer, Samuel S Schoenholz, Patrick F Riley, Oriol Vinyals, and George E Dahl. 2017. Neural message passing for quantum chemistry. In ICML. PMLR, 1263--1272.Google Scholar
- Robert Glem, Andreas Bender, Catrin Hasselgren, Lars Carlsson, Scott Boyer, and James Smith. 2006. Circular fingerprints: Flexible molecular descriptors with applications from physical chemistry to ADME. IDrugs : The Investigational Drugs Journal, Vol. 9 (2006), 199--204.Google Scholar
- Alexander Grigor'yan. 2009. Heat Kernel and Analysis on Manifolds .American Mathematical Society.Google Scholar
- Christoph Hofer, Roland Kwitt, Marc Niethammer, and Andreas Uhl. 2017. Deep Learning with Topological Signatures. In NeurIPS, I. Guyon, U. V. Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett (Eds.). Curran Associates, Inc., 1634--1644.Google Scholar
- Christoph D. Hofer, Florian Graf, Bastian Rieck, Marc Niethammer, and Roland Kwitt. 2020. Graph Filtration Learning. In ICML, Hal Daumé III and Aarti Singh (Eds.), Vol. 119. PMLR, 4314----4323.Google Scholar
- Thomas N. Kipf and Max Welling. 2017. Semi-Supervised Classification with Graph Convolutional Networks. In ICLR .Google Scholar
- Risi Kondor and Horace Pan. 2016. The Multiscale Laplacian Graph Kernel. In NeurIPS, D. D. Lee, M. Sugiyama, U. V. Luxburg, I. Guyon, and R. Garnett (Eds.). Curran Associates, Inc., 2990--2998.Google Scholar
- Nils Kriege and Petra Mutzel. 2012. Subgraph Matching Kernels for Attributed Graphs. In ICML .Google Scholar
- Nils M. Kriege, Pierre-Louis Giscard, and Richard C. Wilson. 2016. On Valid Optimal Assignment Kernels and Applications to Graph Classification. In NeurIPS. 1623--1631.Google Scholar
- Nils M. Kriege, Fredrik D. Johansson, and Christopher Morris. 2020. A Survey on Graph Kernels. Applied Network Science, Vol. 5, 1 (2020), 6.Google ScholarCross Ref
- Yong Lin, Linyuan Lu, and Shing-Tung Yau. 2011. Ricci curvature of graphs. Tohoku Mathematical Journal, Second Series, Vol. 63, 4 (2011), 605--627.Google ScholarCross Ref
- H. Ling and K. Okada. 2007. An Efficient Earth Mover's Distance Algorithm for Robust Histogram Comparison. IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 29, 5 (2007), 840--853.Google ScholarDigital Library
- Christopher Morris, Nils M. Kriege, Kristian Kersting, and Petra Mutzel. 2016. Faster Kernels for Graphs with Continuous Attributes via Hashing. In Proceedings of the 16th IEEE International Conference on Data Mining. 1095--1100.Google ScholarCross Ref
- T. Ramraj and R. Prabhakar. 2015. Frequent Subgraph Mining Algorithms -- A Survey. Procedia Computer Science, Vol. 47 (2015), 197--204. Graph Algorithms, High Performance Implementations and Its Applications ( ICGHIA 2014 ).Google ScholarCross Ref
- Bastian Rieck, Christian Bock, and Karsten Borgwardt. 2019. A Persistent Weisfeiler--Lehman Procedure for Graph Classification. In ICML. PMLR, 5448--5458.Google Scholar
- Bastian Rieck, Filip Sadlo, and Heike Leitte. 2020. Topological Machine Learning with Persistence Indicator Functions. In Topological Methods in Data Analysis and Visualization V, Hamish Carr, Issei Fujishiro, Filip Sadlo, and Shigeo Takahashi (Eds.). Springer, 87--101.Google Scholar
- Kaspar Riesen. 2015. Graph Edit Distance. In Structural Pattern Recognition with Graph Edit Distance: Approximation Algorithms and Applications. Springer, Chapter 2, 29--44.Google Scholar
- Nino Shervashidze and Karsten Borgwardt. 2009. Fast subtree kernels on graphs. In NeurIPS. 1660--1668.Google Scholar
- N. Shervashidze, P. Schweitzer, E. Jan van Leeuwen, K. Mehlhorn, and K. M. Borgwardt. 2011. Weisfeiler--Lehman Graph Kernels. Journal of Machine Learning Research 12 (2011), 2539--2561.Google ScholarDigital Library
- Martin Simonovsky and Nikos Komodakis. 2017. Dynamic edge-conditioned filters in convolutional neural networks on graphs. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 3693--3702.Google ScholarCross Ref
- Jeffrey J. Sutherland, Lee A. O'Brien, and Donald F. Weaver. 2003. Spline-fitting with a genetic algorithm: a method for developing classification structure-activity relationships. Journal of Chemical Information and Computer Sciences, Vol. 43, 6 (2003), 1906--1915.Google ScholarCross Ref
- Matteo Togninalli, Elisabetta Ghisu, Felipe Llinares-López, Bastian Rieck, and Karsten Borgwardt. 2019. Wasserstein Weisfeiler--Lehman Graph Kernels. In NeurIPS, H. Wallach, H. Larochelle, A. Beygelzimer, F. d'Alché Buc, E. Fox, and R. Garnett (Eds.). Curran Associates, Inc., 6436--6446.Google Scholar
- Anton Tsitsulin, Davide Mottin, Panagiotis Karras, Alexander Bronstein, and Emmanuel Müller. 2018. NetLSD: Hearing the Shape of a Graph. In Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. 2347--2356.Google ScholarDigital Library
- Yuhei Umeda. 2017. Time Series Classification via Topological Data Analysis. Transactions of the Japanese Society for Artificial Intelligence, Vol. 32, 3 (2017), D--G72_1--12.Google ScholarCross Ref
- Zonghan Wu, Shirui Pan, Fengwen Chen, Guodong Long, Chengqi Zhang, and Philip S. Yu. 2021. A Comprehensive Survey on Graph Neural Networks. IEEE Transactions on Neural Networks and Learning Systems, Vol. 32, 1 (2021), 4--24.Google Scholar
- Keyulu Xu, Weihua Hu, Jure Leskovec, and Stefanie Jegelka. 2019. How Powerful are Graph Neural Networks?. In ICLR .Google Scholar
- Ze Ye, Kin Sum Liu, Tengfei Ma, Jie Gao, and Chao Chen. 2020. Curvature Graph Network. In ICLR .Google Scholar
- Qi Zhao and Yusu Wang. 2019. Learning metrics for persistence-based summaries and applications for graph classification. In NeurIPS. 9855--9866.Google Scholar
Index Terms
- Filtration Curves for Graph Representation
Recommendations
Hierarchical Graph Representation Learning with Structural Attention for Graph Classification
Artificial IntelligenceAbstractRecently, graph neural networks (GNNs) exhibit strong expressive power in modeling graph structured data and have been shown to work effectively for graph classification tasks. However, existing GNN models for predicting graph categories ...
An Image-Based Representation for Graph Classification
Structural, Syntactic, and Statistical Pattern RecognitionAbstractThis paper proposes to study the relevance of image representations to perform graph classification. To do so, the adjacency matrix of a given graph is reordered using several matrix reordering algorithms. The resulting matrix is then converted ...
Representation numbers of complete multipartite graphs
A graph G has a representation modulo r if there exists an injective map f:V(G)->{0,1,...,r-1} such that vertices u and v are adjacent if and only if f(u)-f(v) is relatively prime to r. The representation number rep(G) is the smallest r such that G has ...
Comments