Training Files

For networks that use supervised learning schemes, desired values must be specified for each input value (e.g. LMS Networks and Backprop Networks). These are specified using simple lists of comma-separated-values that are stored in .csv files. One file should exist for input values, while another should exist for output values, both with the same number of rows.

Example:

To teach a network the following set of associations,

(0,0) > (0)
(1,0) > (1)
(0,1) > (1)
(1,1) > (0)

one would create two files. Let us call them input.csv and output.csv.

input.csv

0,0
1,0
0,1
1,1

output.csv

0
1
1
0

This corresponds to the standard exclusive or function.