This repository contains the codes of Pseudo Random Fourier Features for Approximating RBF kernels in a data dependent manner. The potential of this method is demonstrated in the classification and regression tasks.
For details of this method, please refer to: Data Dependent Kernel Approximation using Pseudo Random Fourier Features
File to run:
MainFile_OptPRFF_RFF_Nyst_ORF.py
-
this file includes RFF, PRFF (proposed), ORF, and Nystrom method
-
for the Psudeo random fourier features, please have a look at:prff_bharath.py and prff.py(variant)
For the ORF, you need to install the 'revrand' library