Posts Tagged ‘Random Features’

Ali Rahimi and Ben Recht:
Random Features for Large Scale Kernel Machines
NIPS 2007

In this paper, the authors propose to map data to a low-dimensional Euclidean space, such that the inner product in this space is a close approximation of the inner product computed by a stationary (shift-invariant) kernel (in a potentially infinite-dimensional RKHS). The approach is based on Bochner’s theorem.

The central equation is this one: (more…)