


MLPINIT(1)               USER COMMANDS                 MLPINIT(1)



NAME
     mlpinit - generate MLP configuration file from training data

SYNOPSIS
     mlpinit (-u|-h numhidden) patbase > model.mlp

DESCRIPTION
     mlpinit creates an approximation to  an  MLP  matched  to  a
     given  set  of  training data.  This may then be used as the
     starting point for back-propagation training.   The  default
     action  is  to produce a linear network (i.e. with no hidden
     units) with the number of input and output  found  from  the
     training  data.  If the -h switch is used, mlpinit creates a
     two layer network, with a hidden layer comprising  at  least
     as many units as specified (but possibly more).

     Options:

     -I         Identify the program name and version.

     -hnumhidden
                Specify the minimum number of units in the hidden
                layer.  Default 0.

     -u         Output an unlayered network with no hidden units.

VERSION/AUTHOR
     1.1 - Mark Huckvale

SOURCE
     /users/mark/sfs/man/../prw/mlpw.src/mlpinit.c
























PRW                     Last change: UCL                        1



