1 * This data set is designed to show how associative memory works.
\r
2 * The characteristics of a number of people and their names are stored
\r
5 * #1 a person from Chicago
\r
7 * #3 a person from New York
\r
12 * #8-20 are allocated to names, each position is a unique name
\r
14 * The possible outputs are:
\r
19 * #4 is a Yankees fan
\r
22 * The idea is that you can train these patterns into a backprop net
\r
23 * and then test the net with certain inputs. For instance, input
\r
25 * 1010 010 0000000000000
\r
27 * that is, nameless Republican Cub fans from Chicago and you'll get an
\r
28 * estimate of how the person feels about the Sox, Bears, tennis, Yankees
\r
29 * and Jets. Thus, the associative memory does what other people call
\r
30 * fuzzy reasoning but without having to write the fuzzy rules. Actually
\r
31 * the names field in this example could be skipped altogether however
\r
32 * using names has a use in the Hopfield/Boltzman machine so that's how
\r
41 1010 101 1000000000000 01000
\r
42 1010 010 0100000000000 11000
\r
43 0101 101 0010000000000 00001
\r
44 0100 010 0001000000000 00100
\r
45 1000 101 0000100000000 00100
\r
46 0101 011 0000010000000 00001
\r
47 1010 011 0000001000000 11000
\r
48 0101 010 0000000100000 00011
\r
49 1000 000 0000000010000 00100
\r
50 0100 000 0000000001000 00100
\r
51 1010 100 0000000000100 01000
\r
52 0101 100 0000000000010 00001
\r