Random Walk-Based Approach
Explore how random walk techniques create graph embeddings by simulating node sequences. Understand DeepWalk and node2vec algorithms, their parameters, and how they influence embedding quality for graph analytics.
We'll cover the following...
We'll cover the following...
Random walk
As its name suggests, random walk performs a random "walk" on a given graph to compute the embeddings. Take a graph and select a node at random. Now, select its neighbor at random and move to it. This process is repeated until a fixed number of walks is reached. The number of walks is predetermined by the user or tuned as a hyperparameter.
Let's understand this using an example.
Random walk
Consider the example graph above. We fix a maximum walk length of four. We pick a random initial node, say, 1. Then, we randomly ...