CREATING UNDERSTANDABLE MODELS FOR NUMEROUS MODELING TASKS

• US 20150134306A1
• Filed: 11/13/2013
• Published: 05/14/2015
• Est. Priority Date: 11/13/2013
• Status: Abandoned Application

• Current Assignee: GlobalFoundries, Inc.
• Original Assignee: International Business Machines Corporation
• Inventors: Pompey, Pascal, Sinn, Mathieu, Verscheure, Olivier, Wurst, Michael
• Application Number:
  US14/079,170
• Publication Number:
  US 20150134306A1
• Time in Patent Office: Days
• US Class Current: 703/2
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 7/00 Computing arrangements based on specific mathematical models

Abstract

A computer program product for creating models comprises a computer readable storage medium having stored thereon first program instructions executable by a processor to cause the processor to receive the modeling tasks each having a target variable and at least one covariate, the target variable and the at least one covariate being the same for all of the modeling tasks, a relationship between the target variable and the at least one covariate being different for all of the modeling tasks, and second program instructions executable by the processor to cause the processor to generate, for each of the modeling tasks, a model including a transfer function for approximating the relationship between the target value and the at least one covariate of the modeling task in a manner that at least two of the models share an identical transfer function and the models satisfy an accuracy condition.

Claims

1. A computer program product for generating models for a plurality of modeling tasks, the computer program product comprising a computer readable storage medium having stored thereon:

first program instructions executable by a processor to cause the processor to receive the modeling tasks each having a target variable and at least one covariate, the target variable and the at least one covariate being the same for all of the modeling tasks, a relationship between the target variable and the at least one covariate being different for all of the modeling tasks; and

second program instructions executable by the processor to cause the processor to generate, for each of the modeling tasks, a model including a transfer function for approximating the relationship between the target value and the at least one covariate of the modeling task in a manner that at least two of the models share at least one identical transfer function and the models satisfy an accuracy condition.

2. The computer program product of claim 1, wherein the second program instructions comprise:

third program instructions executable by the processor to cause the processor to learn the transfer functions from the modeling tasks such that the transfer functions are different for all of the models;

fourth program instructions executable by the processor to cause the processor to select a subset of the transfer functions; and

fifth program instructions executable by the processor to cause the processor to modify the models by replacing the transfer functions of the models with the subset of the transfer functions.

3. The computer program product of claim 2, wherein the fourth program instructions comprise:

sixth program instructions executable by the processor to cause the processor to group the transfer functions into a plurality of clusters of transfer functions based on similarities of the transfer functions;

seventh program instructions executable by the processor to cause the processor to identify a transfer function in each of the clusters to represents the cluster; and

eighth program instructions executable by the processor to cause the processor to select a set of the representative transfer functions that satisfies the accuracy condition.

4. The computer program product of claim 3, wherein the accuracy condition is satisfied when values approximated by a representative transfer function of a cluster is within a threshold difference from values approximated by a particular transfer function of a model to be replaced by the representative transfer function.

5. The computer program product of claim 2 further comprising third program instructions executable by the processor to cause the processor to forecast target variable values for a particular modeling task using the modified model for the particular modeling task.

6. The computer program product of claim 1 further comprising third program instructions executable by the processor to cause the processor to receive the accuracy condition from a user.

7. The computer program product of claim 1, wherein the second program instructions comprise:

third program instructions executable by the processor to cause the processor to receive, from a user, an input indicating which of the models should share the at least one identical transfer function; and

fourth program instructions executable by the processor to cause the processor to generate the plurality of models based on the input.

8. A system for generating models for a plurality of modeling tasks, the system comprising a processor configured to:

receive the modeling tasks each having a target variable and at least one covariate, the target variable and the at least one covariate being the same for all of the modeling tasks, a relationship between the target variable and the at least one covariate being different for all of the modeling tasks; and

generate, for each of the modeling tasks, a model including a transfer function for approximating the relationship between the target value and the at least one covariate of the modeling task in a manner that at least two of the models share at least one identical transfer function and the models satisfy an accuracy condition.

9. The system of claim 8, wherein the processor is further configured to receive, from a user, an input indicating which of the models should share the at least one identical transfer function and to generate the models based on the input.

10. The system of claim 8, wherein each of the modeling tasks has a data set that includes values of the at least one covariate and values of the target variable, wherein the processor is configured to generate the models further by learning the models simultaneously from the data sets corresponding to the models.

11. The system of claim 10, wherein the learning comprises:

formulating an optimization problem by joining the models;

joining data sets corresponding to the models; and

fitting the models into the joined data sets by solving the optimization problem based on the joined data sets.

12. The system of claim 11, wherein the solving the optimization problem comprises minimizing difference between the target variable values and values approximated by the transfer function.

13. The system of claim 8, wherein the processor is further configured to forecast target variable values for a particular modeling task using the model for the particular modeling task.

14. The system of claim 8, wherein the processor is configured to generate the models by:

learning the transfer functions from the modeling tasks such that the transfer functions are different for all of the models;

selecting a subset of the transfer functions; and

modifying the models by replacing the transfer functions of the models with the subset of the transfer functions.

15-20. -20. (canceled)