SVMs with linear kernels can be trained in O(n) time, as in [LIBLINEAR](https://www.csie.ntu.edu.tw/~cjlin/liblinear/). Can svmTrain fast-path to an O(n) solver when svm.Kernel == "linear"?
SVMs with linear kernels can be trained in O(n) time, as in LIBLINEAR.
Can svmTrain fast-path to an O(n) solver when svm.Kernel == "linear"?