Why Graph?

Graph體現了entities(nodes)之間的聯繫，這些聯繫構成一個network。許多的數據都可以自然的用graph來建模表示，例如計算機網路、疾病傳染路徑、社交網路等。即graph有廣泛的應用空間。

課程討論的主要問題：

How do we take advantage of relational structure for better prediction? 如何利用graph來做更好的預測？

How can we develop neural networks that are much more broadly applicable? 如何將神經網路和graph結合以處理複雜的數據結構？

Modern ML Toolbox

現有的ML工具大多針對簡單的輸入數據結構，如文本語音這種sequence型數據，如圖像這種grid型數據。

Graph representation learning

Representation learning is to map nodes of a graph to a d-dimensional embedding, to d-dimensional vector.

Various topics in ML and Representation learning for graph:

  • Traditional methods: Graphlets, Graph Kernels
  • Methods for node embeddings: Deepwalk, Node2Vec
  • Graph Neural Network: GCN, GraphSAGE,GAT,Theory of GNNs
  • Knowledge graphs and reasoning: TransE, BetaE
  • Applications to Biomedicine Science Industry

參考

  • StandFord CS224W http://web.stanford.edu/class/cs224w/