Efficient Network Representation Learning via Cluster Similarity

Yasuhiro Fujiwara, Yasutoshi Ida, Atsutoshi Kumagai, Masahiro Nakano, Akisato Kimura, Naonori Ueda

Abstract: Network representation learning is a de-facto tool for graph analytics. The mainstream of the previous approaches is to factorize the proximity matrix between nodes. However, if n is the number of nodes, since the size of the proximity matrix is $$n \times n$$ , it needs $$O(n^3)$$ time and $$O(n^2)$$ space to perform network representation learning. The proposed approach computes the representations of the clusters from similarities between clusters and computes the representations of nodes by referring to them. If l is the number of clusters, since $$l \ll n$$ , we can efficiently obtain the representations of clusters from a small $$l \times l$$ similarity matrix. Experiments show that our approach can perform network representation learning more efficiently and effectively than existing approaches.