BiasedWalk: Biased Sampling for Representation Learning on Graphs

TL;DR

BiasedWalk introduces a scalable, unsupervised graph embedding method using biased random walks that adapt between BFS and DFS strategies to effectively capture node similarities, outperforming baseline methods in various tasks.

Contribution

The paper presents BiasedWalk, a novel graph embedding algorithm that employs biased random walks to better capture homophily and role equivalence, improving upon existing methods.

Findings

Outperforms baseline methods on multiple datasets

Effectively captures homophily and role equivalence

Scalable and adaptable to different graph structures

Abstract

Network embedding algorithms are able to learn latent feature representations of nodes, transforming networks into lower dimensional vector representations. Typical key applications, which have effectively been addressed using network embeddings, include link prediction, multilabel classification and community detection. In this paper, we propose BiasedWalk, a scalable, unsupervised feature learning algorithm that is based on biased random walks to sample context information about each node in the network. Our random-walk based sampling can behave as Breath-First-Search (BFS) and Depth-First-Search (DFS) samplings with the goal to capture homophily and role equivalence between the nodes in the network. We have performed a detailed experimental evaluation comparing the performance of the proposed algorithm against various baseline methods, on several datasets and learning tasks. The experiment results show that the proposed method outperforms the baseline ones in most of the tasks and datasets.