Skip to main content
placeholder image

A Trace Ratio Maximization Method for��Parameter Free Multiple Kernel Clustering

Conference Paper


Abstract


  • Multiple Kernel Clustering (MKC) is helpful to leverage complementary information from various contexts and alleviate the difficulty of kernel determination. However, the key weighting strategies for optimal kernel learning around individual kernels are not derived from their optimization problems but embedded in a plug-and-play manner and lead to sub-optimal objective function value. More seriously, the hyper-parameters, introduced by the additive balance of these two coupled sub-tasks, are hard to determine in unsupervised learning scenarios and lead to inconsistent and less satisfying results. To avoid the problems mentioned above, we present a novel parameter-free MKC method with the trace ratio criterion (TRMKC in short), which minimizes the approximation errors between consensus and base kernels using the corr-entropy induced metric and maximizes the mean similarities based on the consensus kernel. The trade-off between these two coupled sub-procedures can be automatically balanced, and the performance could be mutually reinforced. To solve the trace ratio criterion and the corr-entropy induced non-quadratic function optimization problem, we present an alternative strategy with monotonic convergence proof, which reformulates it into a series of sub-problems with trace difference and quadratic programming by utilizing the half-quadratic optimization technique. Extensive MKC experimental results well demonstrate the effectiveness of TRMKC.

Publication Date


  • 2022

Citation


  • Chen, Y., Wang, L., Du, L., & Duan, L. (2022). A Trace Ratio Maximization Method for��Parameter Free Multiple Kernel Clustering. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) Vol. 13246 LNCS (pp. 681-688). doi:10.1007/978-3-031-00126-0_52

Scopus Eid


  • 2-s2.0-85129825556

Web Of Science Accession Number


Start Page


  • 681

End Page


  • 688

Volume


  • 13246 LNCS

Issue


Place Of Publication


Abstract


  • Multiple Kernel Clustering (MKC) is helpful to leverage complementary information from various contexts and alleviate the difficulty of kernel determination. However, the key weighting strategies for optimal kernel learning around individual kernels are not derived from their optimization problems but embedded in a plug-and-play manner and lead to sub-optimal objective function value. More seriously, the hyper-parameters, introduced by the additive balance of these two coupled sub-tasks, are hard to determine in unsupervised learning scenarios and lead to inconsistent and less satisfying results. To avoid the problems mentioned above, we present a novel parameter-free MKC method with the trace ratio criterion (TRMKC in short), which minimizes the approximation errors between consensus and base kernels using the corr-entropy induced metric and maximizes the mean similarities based on the consensus kernel. The trade-off between these two coupled sub-procedures can be automatically balanced, and the performance could be mutually reinforced. To solve the trace ratio criterion and the corr-entropy induced non-quadratic function optimization problem, we present an alternative strategy with monotonic convergence proof, which reformulates it into a series of sub-problems with trace difference and quadratic programming by utilizing the half-quadratic optimization technique. Extensive MKC experimental results well demonstrate the effectiveness of TRMKC.

Publication Date


  • 2022

Citation


  • Chen, Y., Wang, L., Du, L., & Duan, L. (2022). A Trace Ratio Maximization Method for��Parameter Free Multiple Kernel Clustering. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) Vol. 13246 LNCS (pp. 681-688). doi:10.1007/978-3-031-00126-0_52

Scopus Eid


  • 2-s2.0-85129825556

Web Of Science Accession Number


Start Page


  • 681

End Page


  • 688

Volume


  • 13246 LNCS

Issue


Place Of Publication