Pattern recognition system
First Claim
1. A self-organizing pattern recognition system comprising:
- input means for providing a plurality of input elements of an input pattern;
adaptive filter means for individually weighting the input elements relative to each of a plurality of categories;
means for providing, relative to each category, a category selection indication representing a combination of weighted elements from the adaptive filter means and for selecting, based on the category selection indications, at least one selected category;
template means for defining an expected pattern corresponding to the at least one selected category;
means for detecting a sufficient coincidence between the input pattern and the expected pattern; and
means for modifying the adaptive filter means and the expected pattern relative to the at least one selected category where the sufficient coincidence is detected, to retain elements in common with the input pattern and expected pattern and to deemphasize all other elements, and for selecting an alternative category without immediate modification of the adaptive filter means and template means where a sufficient coincidence is not detected, the alternative category then serving to generate, through the template means, an alternative expected pattern to be compared to the input pattern.
1 Assignment
0 Petitions
Accused Products
Abstract
A self-categorizing pattern recognition system includes an adaptive filter for selecting a category in response to an input pattern. A template is then generated in response to the selected category and a coincident pattern indicating the intersection between the expected pattern and the input pattern is generated. The ratio between the number of elements and the coincident pattern to the number of elements in the input pattern determines whether the category is reset. If the category is not reset, the adaptive filter and template may be modified in response to the coincident pattern. Reset of the selected category is inhibited if no expected pattern is generated. Weighting of the adaptive filter in response to a coincident pattern is inversely related to the number of elements in the input pattern. The selected categories reset where a reset function is less than a vigilance parameter which may be varied in response to teaching events.
74 Citations
31 Claims
-
1. A self-organizing pattern recognition system comprising:
-
input means for providing a plurality of input elements of an input pattern; adaptive filter means for individually weighting the input elements relative to each of a plurality of categories; means for providing, relative to each category, a category selection indication representing a combination of weighted elements from the adaptive filter means and for selecting, based on the category selection indications, at least one selected category; template means for defining an expected pattern corresponding to the at least one selected category; means for detecting a sufficient coincidence between the input pattern and the expected pattern; and means for modifying the adaptive filter means and the expected pattern relative to the at least one selected category where the sufficient coincidence is detected, to retain elements in common with the input pattern and expected pattern and to deemphasize all other elements, and for selecting an alternative category without immediate modification of the adaptive filter means and template means where a sufficient coincidence is not detected, the alternative category then serving to generate, through the template means, an alternative expected pattern to be compared to the input pattern. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17)
-
-
18. A self-organizing pattern recognition system comprising:
-
input means for providing a plurality of input elements of an input pattern; adaptive filter means for individually weighting the input elements relative to each of a plurality of categories; means for providing, relative to each category, a category selection indication representing a combination of weighted elements from the adaptive filter means and for selecting, based on the category selection indications, at least one selected category; template means for defining an expected pattern corresponding to the at least one selected category; means for detecting a sufficient coincidence between the input pattern and the expected pattern by weighting a coincidence between the patterns relative to the complexity of the input pattern; means for modifying the adaptive filter means and the expected pattern relative to the at least one selected category, where the sufficient coincidence is detected, and for selecting an alternative category without immediate modification of the adaptive filter means and template means where a sufficient coincidence is not detected, the alternative category then serving to generate, through the template means, an alternative expected pattern to be compared to the input pattern. - View Dependent Claims (19, 20, 21, 22, 23)
-
-
24. A self-organizing pattern recognition system comprising:
-
input means for providing a plurality of input elements of an input pattern; adaptive filter means for individually weighting the input elements relative to each of a plurality of categories; means for providing, relative to each category, a category selection indication representing a combination of weighted elements from the adaptive filter means and for selecting, based on the category selection indications, at least one selected category; template means for defining an expected pattern corresponding to the at least one selected category; means for detecting a sufficient coincidence between the input pattern and the expected pattern by weighting a coincidence between the patterns relative to the complexity of the input pattern; and means for modifying the adaptive filter means and the expected pattern relative to the at least one selected category where the sufficient coincidence is detected, to modify weighting of input elements in common with the input pattern and expected pattern to a degree inversely related to the number of elements in the intersection between the input pattern and the expected pattern and to deemphasize all other elements, and for selecting an alterative category without immediate modification of the adaptive filter means and template means where a sufficient coincidence is not detected, the alternative category then serving to generate, through the template means, an alternative expected pattern to be compared to the input pattern. - View Dependent Claims (25, 26, 27)
-
-
28. In a data processing system, a method of categorizing patterns comprising:
-
storing expected input patterns relative to a plurality of categories; individually weighting input elements of an input pattern relative to each of a plurality of categories and selecting at least one category as an output pattern based on combinations of weighted elements; detecting a sufficient coincidence between the input pattern and a stored expected input pattern corresponding to the at least one selected category of the output pattern; and modifying the weighting of the input elements and modifying the stored expected input pattern relative to the at least one selected category of the output pattern where sufficient coincidence is detected to retain elements in common with the input patterns and expected input pattern and to deemphasize all other elements, and selecting an alternative category without immediate modification of the weighting and expected input pattern where sufficient coincidence is not detected, and repeating the detecting and modifying steps for the alternative category. - View Dependent Claims (29, 30)
-
-
31. In a data processing system, a method of categorizing patterns comprising:
-
storing expected input patterns relative to a plurality of categories; individually weighting input elements of an input pattern relative to each of a plurality of categories and selecting at least one category as an output pattern based on combinations of weighted elements; detecting a sufficient coincidence between the input pattern and a stored expected input pattern corresponding to the at least one selected category of the output pattern by weighting a coincidence between the patterns relative to the complexity of the input pattern; and modifying the weighting of the input elements and modifying the stored expected input pattern relative to the at least one selected category of the output pattern where sufficient coincidence is detected and selecting an alternative category without immediate modification of the weighting and expected input pattern where sufficient coincidence is not detected and repeating the detecting and modifying steps for the alternative category.
-
Specification