RidgeSeeking: Ridge Seeking Algorithm

Description Usage Arguments Value

View source: R/MeanShiftAlgoritms.R

Description

Gradient acent with step size correction according to the direction of the gradient

Usage

1
2
RidgeSeeking(x, h, type = "Gaussian", method = "LHSE", data,
  delta = 1e-06, iter_max = Inf, path = FALSE)

Arguments

x

Starting point [d]

h

Kernel bandwidth parameter

type

Used kernel. Either Gaussian or Epan-(echikov).

method

Used method for the log-density estimation. Either LHSE, LMME or KDE.

data

Data matrix [d, n].

delta

Stopping criteria.

iter_max

Maximal number of iterations before the procedure stops. Will print a warning if it did not converge.

path

TRUE if the whole path of the iteration should be returned.

Value

Fixpoints of the iteration as vector [d]. If path = TRUE it returns a list with

x

Fixpoint [d] of the iteration.

path

Path of the iteration [d, iter], where iter is the number of steps.


Epsilon127/ldder documentation built on March 7, 2020, 7:47 a.m.