Principal Graph Encoder Embedding and Principal Community Detection
Abstract: In this paper, we introduce the concept of principal communities and design a principal graph encoder embedding method to concurrently detect these communities and achieve vertex embedding. Given a graph adjacency matrix with vertex labels, the method computes a sample score for each community, providing a ranking to measure community importance and estimate a set of principal communities. It then produces a vertex embedding by retaining only the dimensions corresponding to the principal communities. We characterize the theoretical properties of the principal graph encoder embedding on the random graph model and prove that the proposed method preserves sufficient information about the vertex labels. The numerical performance of the proposed method is demonstrated through comprehensive simulated and real-data experiments.
Submission Length: Regular submission (no more than 12 pages of main content)
Assigned Action Editor: ~Varun_Kanade1
Submission Number: 2791
Loading