Parallel, multi-unit, adaptive, nonlinear pattern class separator and identifier
First Claim
1. A method for classification and identification of patterns in a system comprising input means for receiving an input signal representing each input pattern as a location in multi-dimensional pattern space, memory means for storing information and computer means connected to said input means and said memory means, said method comprising the steps of:
- (a) storing in said memory means a plurality of prototypes within said multi-dimensional pattern space, each prototype being associated with a particular pattern class;
(b) comparing an input pattern, represented by said signal, with each of said prototypes stored in said memory means to determine whether said input pattern falls within a region of influence of at least one of said prototypes;
(c) storing a new prototype in said memory means when such input pattern fails to fall within the region of influence of any previously stored prototype; and
(d) modifying one or more of said prototypes in said memory means, in response to the receipt of an input pattern, to prevent said input pattern from falling within the sphere of influence of a prototype associated with a different pattern class than said input pattern;
whereby said method tends to develop a pattern space in which each incoming input pattern falls only within the regions of influence of prototypes associated with the same class as the input pattern.
2 Assignments
0 Petitions
Accused Products
Abstract
A system for separating and identifying classes of patterns or events which are not necessarily linearly separable. During a training process of the system, new prototypes are created when prior stored prototypes fail to properly classify an input pattern, and previously stored prototypes are modified when an input pattern falls within the sphere of influence of a prototype associated with a different class than the input pattern.
47 Citations
70 Claims
-
1. A method for classification and identification of patterns in a system comprising input means for receiving an input signal representing each input pattern as a location in multi-dimensional pattern space, memory means for storing information and computer means connected to said input means and said memory means, said method comprising the steps of:
-
(a) storing in said memory means a plurality of prototypes within said multi-dimensional pattern space, each prototype being associated with a particular pattern class; (b) comparing an input pattern, represented by said signal, with each of said prototypes stored in said memory means to determine whether said input pattern falls within a region of influence of at least one of said prototypes; (c) storing a new prototype in said memory means when such input pattern fails to fall within the region of influence of any previously stored prototype; and (d) modifying one or more of said prototypes in said memory means, in response to the receipt of an input pattern, to prevent said input pattern from falling within the sphere of influence of a prototype associated with a different pattern class than said input pattern; whereby said method tends to develop a pattern space in which each incoming input pattern falls only within the regions of influence of prototypes associated with the same class as the input pattern. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9)
-
-
10. In a system for classification and identification of patterns, each pattern being represented by a pattern signal S, comprised of a plurality of signal components s1, s2, . . . sk, said system comprising at least one classification unit Ui including pattern classification means, responsive to said pattern signal S, for producing an output response signal Ri representative of a proposed respective class of said pattern represented by said signal S, said pattern classification means including:
-
(1) memory means for storing a plurality of prototypes within a multi-dimensional pattern space, each prototype including, as prototype information stored in said memory means;
a vector location in said pattern space, a region of influence about said vector location of said prototype, and a particular class with which said prototype is associated;(2) means for comparing the vector location of an input pattern, represented by said pattern signal S, with at least one of said prototype stored in said memory means to determine whether said input pattern location falls within the region of influence of said at least one prototype, and for producing an output response signal Ri, indicative of the class associated with such prototype, if said input pattern location falls within the region of influence thereof; (3) means for storing in said memory means a new prototype, if said input pattern location falls outside the region of influence of said prototypes stored in said memory means, said new prototype including, as prototype information stored in said memory means;
the vector location of said input pattern, a region of influence about said vector location, and the particular class to which said input pattern belongs; and(4) means for modifying the region of influence of a given prototype associated with a particular class if the location of an input pattern of another, different class falls within said region of influence of said given prototype; whereby said system tends to develop a pattern space in which each incoming input pattern falls only within the regions of influence of prototypes associated with the same class as the input pattern. - View Dependent Claims (11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39)
-
-
40. A method for classification and identification of patterns, each pattern being represented by a pattern signal S, comprised of a plurality of signal components s1, . . . s2, . . . sk, said method utilizing a computer system, responsive to said pattern signal S, for producing an output signal R1 representative of a proposed respective class of said pattern represented by said signal S, said computer system comprising (a) memory means, and (b) logic means, coupled to said memory means, for carrying out said method, said method comprising the steps of:
-
(1) storing a plurality of prototypes within a multi-dimensional pattern space, each prototype including, as prototype information stored in said memory means;
a vector location in said pattern space, a region of influence about said vector location of said prototype, and a particular class with which said prototype is associated;(2) comparing the vector location of an input pattern, represented by said pattern signal S, with at least one of said prototypes stored in said memory means to determine whether said input pattern location falls within the region of influence of said at least one prototype, and producing an output response signal Ri, indicative of the class associated with such prototype, if said input pattern location falls within the region of influence thereof; (3) storing in said memory means a new prototype, if said input pattern location falls outside the region of influence of said prototypes stored in said memory means, said new prototype including, as prototype information stored in said memory means;
the vector location of said input pattern, a region of influence about said vector location, and the particular class to which said input pattern belongs; and(4) modifying the region of influence of a given prototype associated with a particular class if the location of an input pattern of another, different class falls within said region of influence of said given prototype; whereby said method tends to develop a pattern space in which each incoming input pattern falls only within the regions of influence of prototypes associated with the same class as the input pattern. - View Dependent Claims (41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69)
-
-
70. A method for classification and identification of patterns in a system comprising input means for receiving an input signal F representing each input pattern as a location in a multi-dimensional pattern space, memory means for storing information and computer means connected to said input means and said memory means, said method comprising the steps of:
-
(a) storing in said memory means a plurality of prototypes within said multi-dimensional pattern space, each prototype including, as prototype information stored in said memory means, a vector location in said multi-dimensional pattern space, a scalar distance defining the size of a region of influence about said vector location in said pattern space, and a particular class with which said prototype is associated; and (b) comparing an input pattern, represented by said signal F, with each of said prototypes stored in said memory means to determine whether said input pattern falls within a region of influence of at least one of said prototypes; wherein said comparing step includes the steps of (1) determining whether the location of said input pattern in said pattern space falls within the region of influence of one or more of said prototypes, and (2) producing a response indicative of the class associated with such prototype if said input pattern falls within the region of influence thereof; wherein each prototype is designated by one of at least two types, said designated type being stored in said memory means for its associated prototype as a part of said prototype information; and wherein a first type of prototype produces a first output signal 1Ri indicative of the certain recognition of the class of an input pattern that falls within the region of influence thereof, and wherein a second type of prototype produces a second output signal 2Ri indicative of a possible, but not certain recognition of the class of an input pattern that falls within the region of influence thereof.
-
Specification