参考:http://scikit-learn.org/stable/modules/kernel_approximation.html
之所以使用approximate explicit feature maps compared to the kernel trick, 是因为这样便于online learning,且能够适用于大数据集。但是还是建议,如果可能,approximate and exact kernel methods应该对比着用。
1、Nystroem Method for Kernel Approximation
Nystroem is a general method for low-rank approximations of kernels. It achieves this by essentially subsampling the data on which the kernel is evaluated. 默认情况下,Nystroem 使用rbf kernel,也可以自定义。 The number of samples used - which is also the dimensionality of the features computed - is given by the parameter