nn.pattern(snns)R Documentation

Create an NN Pattern

Description

nn.pattern creates an object of class nn.pattern from a set of input, output and target output matrices. Note that (some norm of) the difference between target and output equals the network output error.

Neural network classifiers cannot be trained on class labels, but on encoded versions thereof. Class nn.pattern supports this by providing two list elements for output and target.

Usage

nn.pattern(input, target=NULL, output=NULL, code=NULL, target.labels=NULL, output.labels=NULL)

Arguments

input matrix with each row containing one input vector
target matrix with each row containing one teaching output vector
output matrix with each row containing one output vector
code matrix with each row containing the encoding of a class
target.labels vector containing target class labels
output.labels vector containing output class labels


[Package Contents]