Watch Your Step: Learning Node Embeddings via Graph Attention

By Sami Abu-El-Haija, Bryan Perozzi, Rami Al-Rfou, Alexander A. Alemi. 2017

207 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

Predict then Propagate: Graph Neural Networks meet Personalized PageRank (2018, methodology): “N-GCN uses h = 16 hidden units, R = 4 heads per random walk length and random walks of up to K − 1 = 4 steps.”
Graph Learning: A Survey (2021, background): “Attention Walks (AWs) are another kind of learning model based on GNN and random walks [146].”
An Attentive Survey of Attention Models (2019, background): “[2018], Abu-El-Haija et al. [2018], Feng et al. [2016] Hyperedge Detection Zhang et al.”
Machine Learning on Graphs: A Model and Comprehensive Taxonomy (2020, methodology): “WYS shows that, when viewed in expectation, these hyperparameters only correspond in the objective to coefficients to the powers of the adjacency matrix (W k)1≤k≤K .”
Diffusion Improves Graph Learning (2019, methodology): “We also tried obtaining θk from models that learn analogous coefficients [1, 8, 13].”
Attention Models in Graphs (2018, methodology): “This can be for different reasons, for instance [Abu-El-Haija et al. 2018] uses attention to bias the walk sampling process which in turn has an effect on the learned node embeddings.”
Adaptive Universal Generalized PageRank Graph Neural Network (2020, background): “Some prior work also emphasizes adaptively learning the importance of different steps (Abu-El-Haija et al., 2018; Berberidis et al., 2018).”
Multi-scale Attributed Node Embedding (2019, methodology): “h lengths (or scales) equally or according to ﬁxed weightings (Mikolov et al., 2013a; Grover & Leskovec, 2016); whereas it has been found that an optimal weighting may be task or dataset speciﬁc (Abu-El-Haija et al., 2018). In contrast, multi-scale node embedding methods, such as LINE (Tang et a”
An Overview of Deep Semi-Supervised Learning (2020, methodology): “Another family of method is random walks introduced by [117] and its variants [57, 47, 20, 1].”
AWB-GCN: A Graph Convolutional Network Accelerator with Runtime Workload Rebalancing (2019, methodology): “ address problems in graph processing. The ﬁrst GNN model was proposed by Gori et al. [20]. In the past decade, work has continued on optimizing GNN algorithms exploring new neural network approaches [2,14,18,33,38,39,45]. More recently, inspired by CNNs that achieve great success with euclidean dat”