Efficient Multiple Kernel Learning Algorithms Using Low-Rank Representation

Abstract

Unlike Support Vector Machine (SVM), Multiple Kernel Learning (MKL) allows datasets to be free to choose the useful kernels based on their distribution characteristics rather than a precise one. It has been shown in the literature that MKL holds superior recognition accuracy compared with SVM, however, at the expense of time consuming computations. This creates analytical and computational difficulties in solving MKL algorithms. To overcome this issue, we first develop a novel kernel approximation approach for MKL and then propose an efficient Low-Rank MKL (LR-MKL) algorithm by using the Low-Rank Representation (LRR). It is well-acknowledged that LRR can reduce dimension while retaining the data features under a global low-rank constraint. Furthermore, we redesign the binary-class MKL as the multiclass MKL based on pairwise strategy. Finally, the recognition effect and efficiency of LR-MKL are verified on the datasets Yale, ORL, LSVT, and Digit. Experimental results show that the proposed LR-MKL algorithm is an efficient kernel weights allocation method in MKL and boosts the performance of MKL largely.

Feature Selection and Combination of Information in the Functional Brain Connectome for Discrimination of Mild Cognitive Impairment and Analyses of Altered Brain Patterns

“…To overcome these challenges, we employed MKL-SVM for information combination. MKL (Niu et al, 2017) is a sparse machinelearning method that allows identification of the most relevant classification sources. The results suggested that the performance of classification by combining multiple brain connectome features was better than that of individual connectome features.…”

“…Therefore, the multimodal brain network (i.e., functional connections and graph theory topological metrics) should be used to provide a comprehensive and insightful understanding of the brain network in patients with MCI. Combined with information from different attributes, multiple kernel learning SVM (MKL-SVM) (Niu et al, 2017) can partially alleviate the high-dimensional curve of multiple features and measure the contributions of different features to the classification. These proposed methods could help select critical features and discriminate normal controls from subjects with diseases.…”

Fixed-budget approximation of the inverse kernel matrix for identification of nonlinear dynamic processes

“…As for a greedy selection procedure, it requires at least O(MN 2 ) additional operations for calculation of a selected criterion, which is likely to be too expensive. The second approach is based on reduced-rank approximations of the kernel matrix [13][14][15][16].…”