On the Impact of Kernel Approximation on Learning Accuracy
Kernel approximation is commonly used to scale kernel-based algorithms to applications containing as many as several million instances. This paper analyzes the effect of such approximations in the ...