Next: Functionality
Up: Artificial Neural Networks Simulation,
Previous: Directory Hierarchy
Usage
To be able to train a network using ANNAP, the generic network definition file
Base/ProbSpec.m6 should be customized. This customized file then should be placed in a directory
under the root directory of ANNAP, preferably named by the problem that it tries
to solve.
The data set files should also be placed in the same directory. The data set
is represented by two files in ASCII format and every line in these files consist
of white space separated input values for a single pattern. Therefore, number
of lines in both the files will be the same and will also be equal to number
of patterns to present to the network.
- Input file
- The file that consist of lines representing the input patterns to
the network (see Section 4.1.3 for implementation details).
- Output file
- The file that consist of lines representing the output activations
of the network stimulated by the corresponding input patterns. Note that this
is still an input file to be read by the program and loaded as the desired outputs
of the network (see Section 4.1.3 for implementation details).
Next: Functionality
Up: Artificial Neural Networks Simulation,
Previous: Directory Hierarchy
Cengiz Gunay
2000-06-25