On the Equivalence between Positional Node Embeddings and Structural Graph Representations

By Balasubramaniam Srinivasan, Bruno Ribeiro. 2019

143 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.

In Graph Learning

Citations in the corpus

E(n) Equivariant Graph Neural Networks (2021, background): “Another way of looking at this is considering the sampled noise makes the node representations go from structural to positional (Srinivasan & Ribeiro, 2019) where E( n ) equivariance may be beneﬁcial.”
Inductive Representation Learning in Temporal Networks via Causal Anonymous Walks (2021, background): “Note that GraphSAGE (Hamilton et al., 2017a) and GAT (Veliˇckovi´c et al., 2018) also share the similar issue when representing static networks for link prediction (Zhang et al., 2020; Srinivasan & Ribeiro, 2019).”
GraphSAIL: Graph Structure Aware Incremental Learning for Recommender Systems (2020, background): “it fails to capture each node’s global position information, which is the relative position of the node with respect to all the other nodes. This is a general limitation of the majority of GNN models [21, 30]. However, each node’s global position encodes rich information for the recommendation setti”
How Powerful are Spectral Graph Neural Networks (2022, unknown): “The node labels assigned by 1-WL test can also be used to check if two nodes are isomorphic (Babai & Kucera, 1979; Srinivasan & Ribeiro, 2020), as two isomorphic nodes always have the same label while two non-isomorphic ones mostly have different labels.”
Graph Neural Networks with Learnable Structural and Positional Representations (2021, methodology): “Given the recent theoretical results on the importance of informative graph PE for expressive GNNs (Murphy et al., 2019; Srinivasan & Ribeiro, 2019; Loukas, 2020), we are interested in a generic framework that can enable GNNs to separate positional and structural representations to increase their…”
Machine Learning on Graphs: A Model and Comprehensive Taxonomy (2020, background): “While some models are inherently better suited to different tasks in practice, recent theoretical results (Srinivasan and Ribeiro, 2020) show that models previously assumed to be capable of only one setting (e.”