The files contained in this directory have the following significance :

- pattern_src
	a set of "grid" files corresponding to each of the patterns in
	tc.pat (see descriptive section below for more details)

- tc.pat
	a set of pattern files (see descriptive section below for more
	details)

- tc.readme
 	this file


**NOTE :  training on this pattern set will likely take many epochs
	  (the best configurations i've tried required ~400)


T-C Description
---------------

The pattern file 'tc.pat' contains 56 81-input, 2-output patterns.
The inputs of each pattern correspond to a 9X9 "grid" of bits -- the
network's input vector for each grid is just the grid's elements in
row major order starting from the top of the grid.   The task of the 
network is to detect any number of 5-bit T and/or C patterns on an 
input grid.   For example, the following grids both contain occurrences
of both T and C patterns :

	Grid 1:                       Grid 2:	
	0 0 0 0 0 0 0 0 0             1 1 1 1 1 0 0 0 1 
	0 1 1 1 0 0 0 1 0             1 0 1 1 1 0 1 0 1 
	0 0 1 0 0 0 0 0 1             1 0 1 1 1 1 1 1 1 
	0 0 1 0 0 1 0 1 0             0 0 0 1 1 1 1 1 1 
	0 0 0 0 0 0 1 0 0             1 1 1 1 1 0 1 1 1 
	0 0 0 0 0 0 0 0 0             1 1 1 1 0 1 1 1 1 
	0 0 0 0 1 1 0 0 0             1 1 1 0 1 0 1 1 1 
	0 0 0 0 1 0 0 0 0             1 1 1 1 1 1 0 1 1 
	0 0 0 0 1 1 0 0 0             1 1 1 1 1 1 1 1 1 
        
The outputs corresponding to a particular grid are determined as follows :

	Output	|  Grid Pattern Type
	----------------------------
	0 1     |  grid contains one or more C patterns, but no T patterns and
                |     no unrecognized patterns
	1 0     |  grid contains one or more T patterns, but no C patterns and
                |     no unrecognized patterns
	1 1     |  grid contains at least one T and one C pattern, and contains 
                |     no unrecognized patterns
	0 0     |  grid contains no T or C patterns, or contains an unrecognized
		|     pattern 

The only, perhaps, confusing thing about this scheme is that the T or C patterns
may be either represented as a pattern of 1's on a background of 0's, or, 
alternatively, may be represented as a pattern of 0's on a background of 1's.
If either of these interpretations of a particular grid allow the network to
produce an output other than 0 0 (the "rejecting" output), then that 
interpretation is applied -- if neither interpretation results in only T and/or
C patterns, then 0 0 is the output.   In the cases of the example grids shown
above, both Grid 1 and Grid 2 would result in the output 1 1.

Also, notice that the T's and C's can occur at any position and orientation,
so long as they do not "touch" any other T's or C's.   The grid below would
be rejected (produce 0 0 as output) since the two C's it contains "touch" :

	Grid 3:                       
	0 0 0 0 0 0 0 0 0             
	0 1 1 1 0 0 0 1 0             
	0 0 1 0 0 0 0 0 1             
	0 0 1 0 0 1 0 1 0             
	0 0 0 0 0 0 1 0 0             
	0 0 0 0 1 1 0 0 0      
	0 0 0 0 1 0 0 0 0            
	0 0 0 0 1 1 0 0 0        
	0 0 0 0 0 0 0 0 0      


Obviously, not all possible patterns are included in the training set in
tc.pat -- furthermore, I'm not even sure how well a trained network will
generalize to unseen patterns.   Thus, it might be fun/interesting to make up
new patterns and see how well a trained net handles them, or to investigate
whether including them in the training set improves or hinders learning rate
and/or generalization.