Automatic neural-net model generation and maintenance
First Claim
Patent Images
1. A computer-implemented method of incrementally forming and adaptively updating a neural net comprising:
- (a) using a set of sample data patterns to form a hierarchical list of function approximation node candidates, each function approximation node candidate located at the center of a hierarchically arranged cluster;
(b) incrementally adding to the neural net a function approximation node selected from the list of function approximation node candidates;
(c) computing function parameters for the function approximation node and updating function parameters of other nodes in the neural network by using the function parameters of the other nodes prior to addition of the function approximation node to the neural network and(d) storing an updated neural net including the function approximation node and the updated function parameters for use during the recognition of one or more patterns in a new set of data; and
(e) using the updated neural net to improve the performance of a system, wherein the new set of data comprises data that describes a behavior of the system.
1 Assignment
0 Petitions
Accused Products
Abstract
Method of incrementally forming and adaptively updating a neural net model are provided. A function approximation node is incrementally added to the neural net model. Function parameters for the function approximation node are determined and function parameters of other nodes in the neural network model are updated, by using the function parameters of the other nodes prior to addition of the function approximation node to the neural network model.
-
Citations
39 Claims
-
1. A computer-implemented method of incrementally forming and adaptively updating a neural net comprising:
-
(a) using a set of sample data patterns to form a hierarchical list of function approximation node candidates, each function approximation node candidate located at the center of a hierarchically arranged cluster; (b) incrementally adding to the neural net a function approximation node selected from the list of function approximation node candidates; (c) computing function parameters for the function approximation node and updating function parameters of other nodes in the neural network by using the function parameters of the other nodes prior to addition of the function approximation node to the neural network and (d) storing an updated neural net including the function approximation node and the updated function parameters for use during the recognition of one or more patterns in a new set of data; and (e) using the updated neural net to improve the performance of a system, wherein the new set of data comprises data that describes a behavior of the system. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 28, 29, 30)
-
-
25. A computer-implemented method of incrementally forming a supervised learning neural net from data in the form of input-output pairs, comprising:
-
applying a hierarchical clustering methodology to a set of sample data patterns to form a list of function approximation node candidates; incrementally adding one or more function approximation nodes to the supervised learning neural net until the supervised learning neural net has an accuracy level at or above a predetermined accuracy level, wherein the function approximation nodes are selected from the list of function approximation node candidates; and computing function parameters for the function approximation node and updating function parameters of other nodes in the neural network, by using the function parameters of the other nodes prior to addition of the function approximation node to the neural network; storing an updated supervised learning neural net, including the function approximation node and the updated function parameters for use during the recognition of one or more patterns in a new set of data, and using the updated neural net to improve the performance of a system, wherein the new set of data comprises data that describes a behavior of the system. - View Dependent Claims (31, 32, 33)
-
-
26. A computer system, comprising:
- a processor; and
a program storage device readable by the computer system, tangibly embodying a program of instructions executable by the processor to perform a method of incrementally forming and adaptively updating a supervised learning neural net formed from data in the form of input-output pairs, the method comprising;(a) using a set of sample data patterns to form a hierarchical list of function approximation node candidates, each function approximation node candidate located at the center of a hierarchically arranged cluster; (b) incrementally adding to the supervised learning neural net a function approximation node selected from the list of function approximation node candidates; (c) determining function parameters for the function approximation node and updating function parameters of other nodes in the supervised learning neural network, by using the function parameters of the other nodes prior to addition of the function approximation node to the supervised learning neural network; and (d) storing the updated supervised learning neural net including the function approximation node and the updated function parameters for use during the recognition of one or more patterns in a new set of data, and (e) using the updated neural net to improve the performance of a system ˜
wherein the new set of data comprises data that describes a behavior of the system. - View Dependent Claims (34, 35, 36)
- a processor; and
-
27. A program storage device readable by a machine, tangibly embodying a program of instructions executable by the machine to perform a method of incrementally forming and adaptively updating a supervised learning neural net from data in the form of input-output pairs, the method comprising:
-
(a) using a set of sample data patterns to form a hierarchical list of function approximation node candidate, each function approximation node candidate located at the center of a hierarchically arranged clusters; (b) incrementally adding to the supervised learning neural net a function approximation node selected from the list of function approximation node candidates; (c) determining function parameters for the function approximation node and updating function parameters of other nodes in the supervised learning neural network, by using the function parameters of the other nodes prior to addition of the function approximation node to the supervised learning neural network; and (d) storing the updated supervised learning neural net including the function approximation node and the updated function parameters for use during the recognition of one or more patterns in a new set of data; and (e) using the updated neural net to improve the performance of a system, wherein the new set of data comprises data that describes a behavior of the system. - View Dependent Claims (37, 38, 39)
-
Specification