Hybrid neural network generation system and method

• US 20060010089A1
• Filed: 09/02/2005
• Published: 01/12/2006
• Est. Priority Date: 04/06/2001
• Status: Active Grant

• Current Assignee: James Howard Goodnight, John C. Brocklebank, Wolfgang Michael Hartmann
• Original Assignee: James Howard Goodnight, John C. Brocklebank, Wolfgang Michael Hartmann
• Inventors: Brocklebank, John C., Goodnight, James Howard, Hartmann, Wolfgang Michael
• Granted Patent: US 7,162,461 B2
• Time in Patent Office: Days
• US Class Current: 706/21
• CPC Class Codes: G06N 3/045 Combinations of networks
• CPC Class Codes: G06N 3/08 Learning methods

Abstract

A computer-implemented method and system for building a neural network is disclosed. The neural network predicts at least one target based upon predictor variables defined in a state space. First, an input data set is retrieved that includes the predictor variables and at least one target associated with the predictor variables for each observation. In the state space, a number of points is inserted in the state space based upon the values of the predictor variables. The number of points is less than the number of observations. A statistical measure is determined that describes a relationship between the observations and the inserted points. Weights and activation functions of the neural network are determined using the statistical measure.

Claims

1. A computer-implemented method for building an artificial neural network, wherein the artificial neural network predicts at least one target based upon observations defined in a state space, comprising the steps of:

retrieving an input data set that includes the observations and the target;

inserting in the state space a plurality of points based upon the values of the observations in the state space, wherein the number of inserted points is less than the number of observations;

determining a statistical measure that describes a relationship between the observations and the inserted points; and

determining weights and activation functions of the artificial neural network using the statistical measure.