US 7,120,291 B1Granted

Method and apparatus for analyzing input information

OwnerTakafumi Terasawainventor

Life of this patent

Estimating the term…

Loading patent…

Abstract

A method and an apparatus operate like a human neural network to analyze and store input information and form patterns according to the input and stored information. The apparatus has a preprocessing unit (3), an activity computation unit (5), a mutual repression unit (6), and a composition unit (7). The apparatus receives an input pattern, calculates the similarity and activity levels of each stored pattern with respect to the input pattern, and repeats a predetermined number of times the activity calculation of each stored pattern according to the calculated activity level (A(i)), a negative repression coefficient, and the activity levels of the other stored patterns. The apparatus applies final activity levels to cell values of the stored patterns, totals the cell values through the stored patterns, and generates a new pattern according to the totaled cell values.

Claims7

Dependent claims

A method of analyzing input information, comprising the steps of:

converting an input pattern consisting of cells into a converted pattern consisting of cells and storing the converted pattern;
calculating an activity level of a partial or whole set of cells of at least one of a plurality of stored patterns with respect to the converted pattern according to cell values of the stored patterns and the converted pattern;
repressing each of the calculated activity levels a predetermined number of times according to repression rules that are determined in consideration of a negative repression coefficient and the calculated activity levels of the stored patterns other than the converted pattern;
generating a new cell value for each cell of the at least one stored pattern according to a corresponding one of the cell values and a corresponding one of the repressed calculated activity levels;
and
providing resultant cell values based on the new cell values as a response pattern for the input pattern,
wherein the repressing step changes each activity level A(i) by δA(i) according to a repression coefficient W(i) as follows:
δ
⁢
⁢
A
⁡
(
i
)
=
W
⁡
(
i
)
⁢
{
ψ
⁡
(
A
⁡
(
1
)
)
+
ψ
⁡
(
A
)
⁢
2
)
)
+
…
+
ψ
⁡
(
A
⁡
(
i
-
1
)
)
+
ψ
⁡
(
A
⁡
(
i
+
1
)
)
+
…
+
ψ
⁡
(
A
⁡
(
M
)
)
}
,
and
ψ
⁡
(
x
)
=
{
x
:
x
>
0
0
:
else
.

96 more elements

Dependencies are inferred from the claim text.