nn.pattern(snns) | R Documentation |
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.
nn.pattern(input, target=NULL, output=NULL, code=NULL, target.labels=NULL, output.labels=NULL)
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 |