kernel_nn: Generating A Single Matrix-wise Function Using Neural Network

In statmlhb/CVEK: Cross-Validated Kernel Ensemble

Description

Generate matrix-wise functions for two matrices using neural network kernel.

Usage

kernel_nn(l, d)

Arguments

l	(numeric) A numeric number indicating the hyperparameter (flexibility) of a specific kernel.
d	(integer) For polynomial, d is the power; for matern, v = d + 1 / 2; for rational, alpha = d.

Details

Neural Network Kernels

k_{NN}(x, x')=\frac{2}{π}sin^{-1}\Big(\frac{2\tilde{x}^T \tilde{x}'}{√{(1+2\tilde{x}^T \tilde{x})(1+2\tilde{x}'^T \tilde{x}')}}\Big)

Value

matrix_wise

(function) A function calculating the relevance of two matrices.

Author(s)

Wenying Deng

References

The MIT Press. Gaussian Processes for Machine Learning, 2006.

statmlhb/CVEK documentation built on May 5, 2019, 3:47 a.m.