Practical Neural Network Recipies in C++

406

Chapter 22

Building the Training Set

The training set is built by means of one or more TRAINrfilename

commands. Each time this command is encountered, any existing

training set will be augmented with additional data from the named

file.

To erase the entire training set, use the CLEAR TRAINING

command.

The training file is in ASCII format, with one line per case.

Each line contains one or more real numbers with an optional sign and

decimal point. The number of numbers per line depends on the output

model. The CLASSIFY and AUTO models will have one number for

each input. The GENERAL model will have that, plus one number for

each output. Th ...