Fast Prediction with SVM Models Containing RBF Kernels
This work addresses prediction speed issues for users of SVM models with RBF kernels, offering an incremental improvement through approximation.
The authors tackled the problem of slow prediction times in SVM models with RBF kernels by developing a second-order Maclaurin series approximation, which reduces prediction speed dependency on support vectors to quadratic in input dimensions, resulting in significantly faster predictions and smaller memory footprint when dimensions are few compared to support vectors.
We present an approximation scheme for support vector machine models that use an RBF kernel. A second-order Maclaurin series approximation is used for exponentials of inner products between support vectors and test instances. The approximation is applicable to all kernel methods featuring sums of kernel evaluations and makes no assumptions regarding data normalization. The prediction speed of approximated models no longer relates to the amount of support vectors but is quadratic in terms of the number of input dimensions. If the number of input dimensions is small compared to the amount of support vectors, the approximated model is significantly faster in prediction and has a smaller memory footprint. An optimized C++ implementation was made to assess the gain in prediction speed in a set of practical tests. We additionally provide a method to verify the approximation accuracy, prior to training models or during run-time, to ensure the loss in accuracy remains acceptable and within known bounds.