cems (version 0.4)

cem.example.arc: Conditional Expectation Manifold Example on Arc

Description

This function runs the arc example in:

Samuel Gerber and Ross Whitaker, Conditional Expectation Curves, Submitted 2011.

Usage

cem.example.arc(n=150, noise=0.2, risk=2, sigmaX= 0.1, stepX=0.001,
stepBW=0.01, init = 0, plotEach=1, noiseInit=0.5)

Arguments

n

Sample size.

noise

Amount of normal distributed noise orthogonal to the arc.

risk

Optimization objective

sigmaX

Intial bandwidth for the curve $g$

stepBW

Stepsize for bandwidth optimization

stepX

Stepsize for coordinate optimization

init

Type of initialization. 0 = ground truth, 1 = random, 2 = y-values of arc (i.e. close to principal component)

plotEach

Plot curve after plotEach iterations.

noiseInit

add normal distribution noise to initialization.

References

Samuel Gerber, Tolga Tasdizen, Ross Whitaker, Dimensionality Reduction and Principal Surfaces via Kernel Map Manifolds, In Proceedings of the 2009 International Conference on Computer Vison (ICCV 2009).

Samuel Gerber and Ross Whitaker, Regularization-Free Principal Curve Estimation Journal of Machine Learning Research 2013.

See Also

cem