.png){kind=icon}
Return
K Graph Centers
.png){kind=icon}
PRE
AI
DOI：10.1109/tkde.2026.3701456.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
K-means is a powerful unsupervised clustering method for regularly structured data, relying fundamentally on geometric cluster centers. However, extending this paradigm to graph-structured data remains challenging. We propose ”K-Graph-Centers” — a novel method that defines the graph cluster via ”graph center”, represented as an $N$-dimensional vector minimizing the sum of weighted topological distances within its cluster. We introduce an efficient power iteration scheme to compute graph centers by: 1) diffusing affinity through normalized adjacency matrix; 2) enforcing sparsity via row-wise discretization. Cluster assignments are derived from the converged graph centers, similar with K-means. K-Graph-Centers method exhibits rather lower computational complexity and fast convergence speed, compared to traditional spectral-based methods. The experimental results verify its time efficiency and clustering performance in common metrics.
Keywords:
K-graph-centers
virtual graph center
spectral clustering
graph cohesiveness
K-means
Journal
IF:
10.4
Papers:
6.8K
Citations:
3.2W
Organization
Cited Papers
Clustering and projected clustering with adaptive neighbors.png)
Proceedings of the 20th ACM SIGKDD international conference on Knowledge discovery and data mining
Proceedings of the 20th ACM SIGKDD international conference on Knowledge discovery and data miningIF0
PREAI
ShareSaveShareSaveBioMDSum: An Effective Hybrid Biomedical Multi-Document Summarization Method Based on PageRank and Longformer Encoder-DecoderIEEE ACCESS
OAAI
IEEE ACCESSIF3.6
OAAIShareSaveExpert Credibility Prediction Model Based on Fuzzy C-Means Clustering and Similarity Association
PREAI
ShareSave
View more