                  ===============================
                  Description of the tool PAT_SEL
                  ===============================

Given a 'pattern file' and a file containing numbers, PAT_SEL
produces a new 'pattern file' which contains the subset of the first
'pattern file' which consists of the patterns whose numbers are given
in the 'number file'.

Synopsis:
---------

pat_sel <number file> <input pattern file> <output pattern file>

Parameters:
-----------
<number file>		: ASCII file which contains positive integer
			  numbers (one number per line) in ascending
			  order (like it is produced by the tool
			  ANALYZE)

<input pattern file>	: SNNS pattern file

<output pattern file>	: SNNS pattern file which contains the
			  selected subset (created by PAT_SEL)

Examples:
---------

PAT_SEL can be used to create a pattern file which only contains the
patterns which were classified wrong by a neural network. Therefore a
'result file' has to be created using SNNS. The result file can be
analyzed be the tool ANALYZE which produces a file containing the
numbers of wrong classified patterns. This 'number file' and the
corresponding 'pattern file' are read by PAT_SEL and the new 'pattern
file' is going to be created.

