METHODS AND SYSTEM FOR MANAGING PREDICTIVE MODELS

• US 20150347907A1
• Filed: 09/29/2014
• Published: 12/03/2015
• Est. Priority Date: 05/30/2014
• Status: Active Grant

• Current Assignee: Apple Inc.
• Original Assignee: Apple Inc.
• Inventors: WAN, Kit-Man, KAPOOR, Gaurav, MATHEW, Binu K.
• Granted Patent: US 10,380,488 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G06N 5/043 Distributed expert systems; Blackboards
• CPC Class Codes: H04L 67/10 in which an application is distributed across nodes in the network software deployment G06F8/60; multiprogramming arrangements G06F9/46

Abstract

Disclosed herein is a technique for implementing a framework that enables application developers to enhance their applications with dynamic adjustment capabilities. Specifically, the framework, when utilized by an application on a mobile computing device that implements the framework, can enable the application to establish predictive models that can be used to identify meaningful behavioral patterns of an individual who uses the application. In turn, the predictive models can be used to preempt the individual'"'"'s actions and provide an enhanced overall user experience. The framework is configured to interface with other software entities on the mobile computing device that conduct various analyses to identify appropriate times for the application to manage and update its predictive models. Such appropriate times can include, for example, identified periods of time where the individual is not operating the mobile computing device, as well as recognized conditions where power consumption is not a concern.

Claims

1. A method for generating and deploying predictive models to computing devices, the method comprising steps that include:

at a source computing device;

receiving a request to generate a raw predictive model based on a set of training data;

generating the raw predictive model based on the set of training data;

identifying one or more destination computing devices to which the raw predictive model is to be deployed; and

for each of the one or more destination computing devices;

upon determining that it is advantageous to generate a deployment predictive model that is specific to the destination computing device;

generating, using the raw predictive model, the deployment predictive model, and

 

providing the deployment predictive model to the destination computing device.

2. The method of claim 1, wherein the step of determining comprises identifying that a configuration of the raw predictive model can be optimized based on a hardware architecture that is implemented by the destination computing device.

3. The method of claim 1, wherein the step of determining comprises identifying that a configuration of the raw predictive model can be optimized based on a performance requirement instituted by an application that is configured to:

execute on the destination computing device, and

utilize the deployment predictive model.

4. The method of claim 1, wherein each destination computing device of the one or more destination computing devices is configured to implement the provided deployment predictive model.

5. The method of claim 4, wherein each destination computing device of the one or more destination computing devices is configured to gather supplemental training data and periodically modify the provided deployment predictive model based on the supplemental training data.

6. The method of claim 1, wherein each destination computing device of the one or more destination computing devices is a mobile computing device.

7. A method for managing a predictive model on a mobile computing device, the method comprising:

at the mobile computing device;

identifying a condition to perform an update to the predictive model;

accessing an updated set of training data on which to base the update to the predictive model, wherein the updated set of training data supplements an original set of training data previously used to establish the predictive model;

updating the predictive model based on the updated set of training data; and

subsequent to updating the predictive model;

implementing the predictive model for usage on the mobile computing device, and

updating a configuration to reflect that the updated set of training data has been processed.

8. The method of claim 7, wherein identifying the condition comprises receiving a notification from a power management module implemented on the mobile computing device.

9. The method of claim 8, wherein the power management module is configured to provide the notification when the mobile computing device is connected to an external power source and/or when the mobile computing device is not being operated by a user.

10. The method of claim 7, wherein each of the original set of training data and the updated set of training data is stored locally on the mobile computing device.

11. The method of claim 7, wherein the predictive model, subsequent to being updated, is provided to an application that utilizes the predictive model and executes on the mobile computing device.

12. A method for managing a predictive model on a mobile computing device, the method comprising:

at the mobile computing device;

identifying that a threshold amount of additional training data has been added to a set of training data on which the predictive model is based;

determining, based on parameters associated with the predictive model, an appropriate time to update the predictive model;

at the appropriate time, updating the predictive model based on the additional training data; and

subsequent to updating the predictive model, implementing the predictive model for usage.

13. The method of claim 12, wherein the set of training data is stored locally on the mobile computing device.

14. The method of claim 12, wherein the predictive model, subsequent to being updated, is provided to an application that utilizes the predictive model and executes on the mobile computing device.

15. The method of claim 12, wherein the parameters indicate one or more of:

a historical average amount of time required to update to the predictive model, a historical average amount of processing resources that are required to update the predictive model, and a historical average amount of memory resources that are required to update the predictive model.

16. The method of claim 12, wherein the mobile computing device is configured to update at least one other predictive model while the predictive model is being updated.

17. The method of claim 12, further comprising:

receiving, via an over the air update, a replacement predictive model that is associated with the predictive model; and

replacing the predictive model with the replacement predictive model.

18. A non-transitory computer readable storage medium configured to store instructions that, when executed by a processor included in a computing device, cause the computing device to carry out steps that include:

identifying a condition to perform an update to the predictive model;

accessing an updated set of training data on which to base the update to the predictive model, wherein the updated set of training data supplements an original set of training data previously used to establish the predictive model;

updating the predictive model based on the updated set of training data; and

subsequent to updating the predictive model;

implementing the predictive model for usage on the mobile computing device, and

updating a configuration to reflect that the updated set of training data has been processed.

19. The non-transitory computer readable storage medium of claim 18, wherein identifying the condition comprises receiving a notification from a power management module implemented on the mobile computing device.

20. A system, comprising:

a processor; and

a memory configured to store instructions that, when executed by the processor, cause the system to carry out steps that include;

identifying that a threshold amount of additional training data has been added to a set of training data on which the predictive model is based;

determining, based on parameters associated with the predictive model, an appropriate time to update the predictive model;

at the appropriate time, updating the predictive model based on the additional training data; and

subsequent to updating the predictive model, implementing the predictive model for usage.

21. The system of claim 20, wherein the set of training data is stored locally on the mobile computing device.