A Structural Smoothing Framework For Robust Graph Comparison

By Pinar Yanardag, S. V. N. Vishwanathan. 2015

69 citations recorded by the source corpus.

Subfields: Graph Learning, Algorithms & Complexity

How researchers see this paper

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

In Graph Learning

In Algorithms & Complexity

Citations in the corpus

Hierarchical Graph Representation Learning with Differentiable Pooling (2018, methodology): “For D&D and REDDIT-MULTI-12K, the number of hidden dimensions is set to 128.”
subgraph2vec: Learning Distributed Representations of Rooted Sub-graphs from Large Graphs (2016, background): “DOI: 10.1145/1235 lar and widely adopted approaches to measure similarities among graphs [3, 4, 6, 7, 14].”
A simple yet effective baseline for non-attribute graph classification (2018, background): “As pointed out in[YV15b], increasing the size of substructures will largely decrease the probability that two graphs contain similar substructures, which usually results in the “diagonal dominance issue” [KGST03].”
DeepCas: An End-to-end Predictor of Information Cascades (2016, background): “Rather than learning the representation of each node, recent work also attempts to learn the representation of subgraph structures [28, 26, 40, 41].”
Glocalized Weisfeiler-Lehman Graph Kernels: Global-Local Feature Maps of Graphs (2017, methodology): “Subgraph and Weisfeiler-Lehman kernels have been successfully employed within frameworks for smoothed and deep graph kernels [37], [38].”
RetGK: Graph Kernels based on Return Probabilities of Random Walks (2018, background): “As pointed out in [32], increasing the size of substructures will largely decrease the probability that two graphs contain similar substructures, which usually results in the "diagonal dominance issue" [14].”
Relational Pooling for Graph Representations (2019, background): “Graph kernels as special cases of k -ary RP. Graph kernels of Yanardag & Vishwanathan (2015) are special cases of k -ary RP where we assume node and edge attributes are discrete and we hand-pick the f ( A π,π [1: k, 1: k,”
Graph U-Nets (2019, background): “Under inductive learning settings, we compared our meth-ods with other state-of-the-art models on graph classiﬁcation tasks with datasets D&D, PROTEINS, and COLLAB, and the results are summarized in Table 4.”
A survey on graph kernels (2019, methodology): “Similar ideas were used in (Yanardag and Vishwanathan 2015b) where smoothing methods for multinomial distributions were applied to the graph domain.”
Weisfeiler and Leman go sparse: Towards scalable higher-order graph embeddings (2019, methodology): “B INARY datasets.”
Relational Representation Learning for Dynamic (Knowledge) Graphs: A Survey (2019, background): “Common graph classiﬁcation benchmarks include COLLAB [240], PROTEINS [18] and D&D [62].”
Matching Node Embeddings for Graph Similarity (2017, methodology): “The performance of these kernels can be considerably improved by using the Weisfeiler-Lehman test of isomorphism (Shervashidze et al. 2011) or by taking into account similarities between the extracted instances of the considered substructures (Yanardag and Vishwanathan 2015b; 2015a).”