neo(15/5/94) neo(15/5/94)
NAME
neo - a neocognitron simulator.
SYNOPSIS
neo -fin=<fn> -samp=# -fwgt=<fn> -use=# -fspec=<fn>
Example: Let the network read and classify 10 sample pattern
neo -fin=myinput.pat -samp=10 -fwgt=wgts.dat -fspec=specs.dat
-maxiter=100 -reps=5
where
-fin= name of train file containing the input pattern
-samp= number of train input patterns in train file
-fwgt= name of the file with weights (see
documentation)
-use= use the weight file or not (1, 0)
-fspec= name of the file with initial parameters (see
documentation)
-fout= name of the file to write output to (see
documentation)
-maxiter= maximal number of iterations
-reps= number of repetitions for each pattern before
proceeding
DESCRIPTION
A neural network simulator based on Fukushima's Neocognitron.
The Neocognitron is a hierarchical self-organizing feedforward network
. It learns off-line in a supervised or unsupervised manner. The
network was and is mainly used to recognize handwritten numerals and
characters.
AUTHOR
Frank Schnorrenberg
- 1 - Formatted: April 20, 2024
