First, the proposed method offers multiple feature spaces which are designed from GPS data, localization prediction model, image holistic features, and image local features.
Afterwards, we propose a novel method called K-Nearest Neighbors from Multiple Feature Spaces (KNN-MFS) to fuse these candidate sets.
Unlike the traditional Multiple Kernel Learning (MKL) with the implicit kernels, Multiple Empirical Kernel Learning (MEKL) explicitly maps the original data space into multiple feature spaces via different empirical kernels.
Hinrichs et al. [144, 145] and Zhang et al. [4] recently proposed the multi-kernel support vector machine (MK-SVM) algorithm, which is based on multi-kernel learning and extends the kernel tricks in SVM to the multiple feature spaces.
With a focus on the subjects, the feature embedding methods, such as multi-view spectral embedding (MSE) [139] and multi-view local linear embedding (MLLE) [140], have been used to explore the geometric structures of local patches in multiple feature spaces and align the local patches in a unified feature space with maximum preservation of the geometric relationships.
The MKELM method is subsequently developed by integrating these two types of kernels with a multi-kernel learning strategy, which can effectively explore the supplementary information from multiple nonlinear feature spaces for more robust classification of EEG.
Based on affinity propagation (AP) clustering, RReliefF, and sequential forward search, the ARSFS is proposed to select the significant subset of original feature set and to reduce the dimension and multiple correlations of the feature space.
We included the signal from multiple channels into the feature space.
Our framework slices a statistical ADS׳ feature space into multiple disjoint subspaces and then performs anomaly detection separately on each subspace by utilizing more computational resources.
The first approach exploits a small set of features and a single mapping function, while the second approach utilizes a 88-dimensional feature space and multiple distortion-oriented mapping functions.
