Low-rank Kernel Learning for Graph-based Clustering
This work addresses graph-based clustering for data analysis, but it appears incremental as it builds on existing kernel learning approaches.
The paper tackled the problem of graph-based clustering by proposing a low-rank kernel learning method to improve graph construction, resulting in enhanced performance validated through extensive experiments.
Constructing the adjacency graph is fundamental to graph-based clustering. Graph learning in kernel space has shown impressive performance on a number of benchmark data sets. However, its performance is largely determined by the chosen kernel matrix. To address this issue, the previous multiple kernel learning algorithm has been applied to learn an optimal kernel from a group of predefined kernels. This approach might be sensitive to noise and limits the representation ability of the consensus kernel. In contrast to existing methods, we propose to learn a low-rank kernel matrix which exploits the similarity nature of the kernel matrix and seeks an optimal kernel from the neighborhood of candidate kernels. By formulating graph construction and kernel learning in a unified framework, the graph and consensus kernel can be iteratively enhanced by each other. Extensive experimental results validate the efficacy of the proposed method.