graph2kernel: Transform an 'igraph' graph into a p-step random walk kernel

In jdreyf/PANTS: Pathway analysis via network smoothing

Description

Transforms an igraph input graph into a p-step random walk kernel, using parameters a and p. This is similar to netClass::calc.diffusionKernelp, except that function takes slightly different input. The equation for the kernel is (a * I - L)^p. Up to scaling terms, this is equivalent to a p-step random walk on the graph with random restarts, so it is similar to the diffusion kernel, but can be calculated more cheaply (Smola & Kondor).

Usage

graph2kernel(gr, a = 2, p = 1)

Arguments

gr	graph object of class igraph.
a	parameter of p-step random walk kernel, must be >= 2.
p	parameter of p-step random walk kernel, must be > 0.

Value

Laplacian kernel matrix.

References

Smola and Kondor, "Kernels and Regularization on Graphs" In Learning Theory and Kernel Machines, Vol. 2777 (2003), pp. 144-158, doi:10.1007/978-3-540-45167-9_12.

jdreyf/PANTS documentation built on July 18, 2019, 10:12 a.m.