Machine Learning on Graphs: A Model and Comprehensive Taxonomy
Citations
2,494 citations
1,266 citations
481 citations
353 citations
346 citations
References
72,897 citations
"Machine Learning on Graphs: A Model..." refers methods in this paper
...[17] introduced NGM and showed that the regularization objective in Equation 26 generalizes to more complex neural architectures than feed-forward neural networks (FF-NN), such as Long short-term memory (LSTM) networks [54] or CNNs [68]....
[...]
52,856 citations
49,639 citations
"Machine Learning on Graphs: A Model..." refers background in this paper
...with one terabyte size) but does not fit on RAM, how can a researcher apply a learning method on such a large graphs, using just a personal computer? Contrast this with a computer vision task by considering a large image dataset [33, 67]....
[...]
30,558 citations
"Machine Learning on Graphs: A Model..." refers methods in this paper
...4 Outer product-based: Skip-gram methods Skip-gram graph embedding models were inspired by efficient NLP methods modeling probability distributions over words for learning word embeddings [80, 89]....
[...]
30,124 citations
"Machine Learning on Graphs: A Model..." refers methods in this paper
...Note that unsupervised graph embedding methods are also used for visualization purposes: by first training an encoder-decoder model (corresponding to a shallow embedding or graph convolution network), and then mapping every node representation onto a two-dimensional space using, t-distributed stochastic neighbor embeddings (t-SNE) [77] or PCA [58]....
[...]