The AddNoise node in SciCraft is very simple and has one input port
and two input parameters:
- Input port: X which is the input data matrix. Each row is
an object and each column contains the different variables
(features). This data matrix will be added noise
- Input parameter Method which can take two states:
Flat and Normal. The Method parameter is related to
the shape of the distribution which can be either flat or like a
Gaussian curve.
- Input parameter level which is a real number indicating
how much the random contribution is.
The node has two output variables:
- Out which is the original data matrix X added noise
- SNR which is a scalar which contains the estimate of the
signal to noise ratio of the data in the Out matrix.
It is planned that this node will later be expanded to handle
heteroscedastic noise and other types of noise distributions.
Bjørn Kåre Alsberg
2006-04-06