@article {IOPORT.06001652,
    author       = {Wu, Zhengpeng and Zhang, Xuegong},
    title        = {Elastic multiple kernel learning.},
    year         = {2011},
    journal      = {Acta Automatica Sinica},
    volume       = {37},
    number       = {6},
    issn         = {0254-4156},
    pages        = {693-699},
    publisher    = {Science Press, Beijing},
    doi          = {10.3724/SP.J.1004-2011.00693},
    abstract     = {Summary: Multiple kernel learning (MKL) was presented to deal with kernel fusion. MKL learns a linear combination of several kernels and solves simultaneously the supporting vector machine (SVM) associated with the combined kernel. In the current framework of MKL, sparsity of kernel combination coefficients is encouraged. When a significant portion of the kernels is informative, forcing sparsity tends to select only a few kernels and may ignore useful information. In this paper, we propose elastic multiple kernel learning (EMKL) to achieve adaptive kernel fusion. EMKL makes use of a mixing regularization function to compromise sparsity and non-sparsity. Both MKL and SVM could be regarded as special cases of EMKL. Based on the gradient descent algorithm for the MKL problem, we propose a fast algorithm to solve the EMKL problem. Results on simulation datasets demonstrate that the performance of EMKL compares favorably to both MKL and SVM. We further apply EMKL to gene set analysis and get promising results. Finally, we study the theoretical advantages of EMKL comparing it to non-sparse MKL.},
    identifier   = {06001652},
}