00010		For those not familiar with the use of signature tables as
00020	used by Samuel in programs which played the game of checkers, the
00030	concept is best illustrated (Fig.1) by an arrangement of tables
00040	used in the program. There are 27 input terms. Each term evaluates
00050	a specific aspect of a board situation and it is quantized into
00060	a limited but adequate range of values, 7,5,and 3, in this case.
00070	The terms are divided into 9 sets with 3 terms each, forming the 9
00080	first level tables. Outputs from the first level tables are quantized
00090	to 5 levels and combined into 3 second level tables and, finally,
00100	into one third-level table whose output represents the figure of
00110	merit of the board in question.
00120		A signature table has an entry for every possible combination
00130	of the input vector. Thus there are 7*5*3 or 105 entries in each of
00140	the first level tables. Training consists of accumulating two counts
00150	for each entry during a training sequence. Count A is incremented
00160	when the current input vector represents a prefered move and count D
00170	is incremented when it is not the prefered move. The output from the
00180	table is computed as a correlation coeficient
00190				C=(A-D)/(A+D)
00200	The figure of merit for a board is simply the coefficient obtained
00210	as the output from the final table.
00220	
00230		The following three advantages emerge from this method of
00240	training and evaluation.
00250		1) Essentially arbitrary inter-relationships between the input
00260	terms are taken in account by any one table. The only loss of
00270	accuracy is in the quantization.
00280		2) The training is a very simple process of accumulating
00290	counts. The training samples are introduced sequentially, and hence
00300	simultaneous storage of all the samples is not required.
00310		3) The process linearizes the storage requirements in the
00320	parameter space. In the case shown this requires only 343 entries
00330	instead of the 105↑9 entries were the entire space to be represented.
00340