US 2022/0092346 A1Published application

Computer-based systems, computing components and computing objects configured to implement dynamic outlier bias reduction in machine learning models

AssigneeHartford Steam Boiler Inspection and Insurance Company

InventorRICHARD B. JONES

Life of this patent

Estimating the term…

Loading patent…

Abstract

Systems and methods include processors for receiving training data for a user activity; receiving bias criteria; determining a set of model parameters for a machine learning model including: (1) applying the machine learning model to the training data; (2) generating model prediction errors; (3) generating a data selection vector to identify non-outlier target variables based on the model prediction errors; (4) utilizing the data selection vector to generate a non-outlier data set; (5) determining updated model parameters based on the non-outlier data set; and (6) repeating steps (1)-(5) until a censoring performance termination criterion is satisfied; training classifier model parameters for an outlier classifier machine learning model; applying the outlier classifier machine learning model to activity-related data to determine non-outlier activity-related data; and applying the machine learning model to the non-outlier activity-related data to predict future activity-related attributes for the user activity.

Claims20

Dependent claims

A method comprising:

receiving, by at least one processor from at least one computing device associated with at least one production environment, a production-ready model request comprising a training data set of data records, having actual values;
wherein each data record comprising an independent variable and a target variable;
wherein each data record comprises an actual value associated with the target variable;
determining, by the at least one processor, at least one bias criteria;
selecting, by the at least one processor, at least one machine learning model based at least in part on the production-ready model request;
determining, by the at least one processor, a set of model parameters for the at least one machine learning model, comprising:
(1) applying, by the at least one processor, the at least one machine learning model having a set of initial model parameters to the training data set to determine a set of model predicted values;
(2) generating, by the at least one processor, an error set of data element errors by comparing each model predicted value in the set of model predicted values to a corresponding actual value of the actual values of each data record in the training data set;
(3) generating, by the at least one processor, a data selection vector to identify non-outlier target variables based at least in part on the error set of data element errors and the at least one bias criteria;
(4) applying, by the at least one processor, the data selection vector to the training data set to generate a non-outlier data set;
(5) determining, by the at least one processor, a set of updated model parameters for the at least one machine learning model based on the non-outlier data set;
and
(6) repeating, by the at least one processor, at least one iteration of steps (1)-(5) until at least one censoring performance termination criterion is satisfied so as to obtain the set of model parameters for the at least one machine learning model as the updated model parameters, whereby each iteration re-generates the set of predicted values, the error set, the data selection vector, and the non-outlier data set using the set of updated model parameters as the set of initial model parameters;
determining, by the at least one processor, a production-ready machine learning model of the at least one machine learning model based at least in part on a total error of the error set of data element errors;
and
transmitting, by the at least one processor, the production-ready machine learning model to the at least one computing device for use in the at least one production environment.

13 more elements

A method comprising:

transmitting, by at least one processor of at least one computing device associated with at least one production environment, a production-ready model request comprising a training data set of data records to at least one automated model generation processor;
wherein each data record comprising an independent variable and a target variable;
wherein each data record comprises an actual value associated with the target variable;
receiving, by the at least one processor from the at least one automated model generation processor, a production-ready machine learning model, having actual values;
wherein the production-ready machine learning model is selected from at least one machine learning model based at least in part on a total error of an error set of data element errors;
wherein the total error is based at least in part on at least one iteration performed by the at least one automated model generation processor, the at least one iteration comprising:
determining a set of model parameters for the at least one machine learning model comprising:
(1) applying the at least one machine learning model having a set of initial model parameters to the training data set to determine a set of model predicted values;
(2) generating the error set of data element errors by comparing each model predicted value in the set of model predicted values to a corresponding actual value of the actual values of each data record in the training data set;
(3) generating a data selection vector to identify non-outlier target variables based at least in part on the error set of data element errors and at least one bias criteria;
(4) applying the data selection vector to the training data set to generate a non-outlier data set;
(5) determining a set of updated model parameters for the at least one machine learning model based on the non-outlier data set;
and
(6) repeating at least one iteration of steps (1)-(5) until at least one censoring performance termination criterion is satisfied so as to obtain the set of model parameters for the at least one machine learning model as the updated model parameters, whereby each iteration re-generates the set of predicted values, the error set, the data selection vector, and the non-outlier data set using the set of updated model parameters as the set of initial model parameters.

11 more elements

A system comprising:

at least one processor configured to execute software instructions that cause the at least one processor to perform steps to:
receive, from at least one computing device associated with at least one production environment, a production-ready model request comprising a training data set of data records, having actual values;
wherein each data record comprising an independent variable and a target variables
wherein each data record comprises an actual value associated with the target variable;
determine at least one bias criteria;
select at least one machine learning model based at least in part on the production-ready model request;
determine a set of model parameters for the at least one machine learning model comprising:
(1) apply the at least one machine learning model having a set of initial model parameters to the training data set to determine a set of model predicted values;
(2) generate an error set of data element errors by comparing each model predicted value in the set of model predicted values to a corresponding actual value of the actual values of each data record in the training data set;
(3) generate a data selection vector to identify non-outlier target variables based at least in part on the error set of data element errors and the at least one bias criteria;
(4) apply the data selection vector to the training data set to generate a non-outlier data set;
(5) determine a set of updated model parameters for the at least one machine learning model based on the non-outlier data set;
and
(6) repeat at least one iteration of steps (1)-(5) until at least one censoring performance termination criterion is satisfied so as to obtain the set of model parameters for the at least one machine learning model as the updated model parameters, whereby each iteration re-generates the set of predicted values, the error set, the data selection vector, and the non-outlier data set using the set of updated model parameters as the set of initial model parameters;
determine a production-ready machine learning model of the at least one machine learning model based at least in part on a total error of the error set of data element errors;
and
transmit the production-ready machine learning model to the at least one computing device for use in the at least one production environment.

14 more elements

A system comprising:

at least one processor of at least one computing device associated with at least one production environment, wherein the at least one processor is configured to execute software instructions that cause the at least one processor to perform steps to:
transmit a production-ready model request comprising a training data set of data records to at least one automated model generation processor;
wherein each data record comprising an independent variable and a target variable;
wherein each data record comprises an actual value associated with the target variable;
receive, from the at least one automated model generation processor, a production-ready machine learning model, having actual values;
wherein the production-ready machine learning model is selected from at least one machine learning model based at least in part on a total error of an error set of data element errors;
wherein the total error is based at least in part on at least one iteration performed by the at least one automated model generation system, the at least one iteration comprising:
determining a set of model parameters for the at least one machine learning model comprising:
(1) applying the at least one machine learning model having a set of initial model parameters to the training data set to determine a set of model predicted values;
(2) generating the error set of data element errors by comparing each model predicted value in the set of model predicted values to a corresponding actual value of the actual values of each data record in the training data set;
(3) generating a data selection vector to identify non-outlier target variables based at least in part on the error set of data element errors and at least one bias criteria;
(4) applying the data selection vector to the training data set to generate a non-outlier data set;
(5) determining a set of updated model parameters for the at least one machine learning model based on the non-outlier data set;
and
(6) repeating at least one iteration of steps (1)-(5) until at least one censoring performance termination criterion is satisfied so as to obtain the set of model parameters for the at least one machine learning model as the updated model parameters, whereby each iteration re-generates the set of predicted values, the error set, the data selection vector, and the non-outlier data set using the set of updated model parameters as the set of initial model parameters.

12 more elements

Dependencies are inferred from the claim text.

Facts

Application: 17/204,940Filed 18 Mar 2021
Priority: 18 Sep 2019
Published: 24 Mar 2022
Term ends: Not estimatedThe source records don’t support an estimate.

Classification16 codes

G06F 18/285
Analysing: selection of pattern recognition techniques, e.g. of classifiers in a multi-classifier system
G06N 20/10
Machine learning: using kernel methods, e.g. support vector machines [SVM]
G06F 30/27
Design optimisation, verification or simulation: using machine learning, e.g. artificial intelligence, neural networks, support vector machines…
G06N 3/09
Learning methods: supervised learning

Family

Application62/902,074Us provisional application · filed 18 Sep 2019
Application17/025,889Continuation in part · filed 18 Sep 2020
Application17/204,940Continuation in part
US 11,288,602 B2Related patent
US 11,328,177 B2Related patent
US 2021/0110313 A1Published application
US 2022/0092346 A1This patent

From the US filing history. Foreign family members aren’t listed.