Research on Theory and Algorithms of Multi-Kernel Graph Clusterings

Abstract

This article introduces the increasingly prominent importance of data analysis and mining due to the explosive growth of data in the information age, particularly in the analysis and mining of graph data. The characteristics of graph data lie in the complex connections between nodes, making its analysis and mining a hot research topic. Traditional clustering algorithms have limitations when dealing with non-linearly separable data, leading to the emergence of multi-kernel graph clustering algorithms. These algorithms utilize multiple kernel functions to compute the similarity between samples in a high-dimensional feature space, thereby better capturing data features and providing more accurate clustering results. The article primarily investigates the principles, algorithms, and applications of various multi-kernel graph clustering algorithms, emphasizing their advantages in handling non-linearly separable data and offering more accurate clustering results. It suggests that further research into these algorithms will enhance clustering algorithm performance and achieve better results in practical applications.