Predictive model importation

• US 8,209,274 B1
• Filed: 10/19/2011
• Issued: 06/26/2012
• Est. Priority Date: 05/09/2011
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Lin, Wei-Hao, Green, Travis H. K., Kaplow, Robert, Fu, Gang, Mann, Gideon S.
• Primary Examiner(s): Holmes, Michael B
• Application Number:
  US13/277,052
• Time in Patent Office: 251 Days
• Field of Search: 706/21
• US Class Current: 706/21
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• 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

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for obtaining a plurality of representations of predictive models, each representation having been received from a different client wherein each representation is associated with a respective user and comprises a description of a respective predictive model, and selecting a model implementation from a plurality of model implementations for each of the obtained representations.

Claims

1. A computer-implemented method, the method comprising:

obtaining a plurality of representations of predictive models, each representation having been received from a different client wherein each representation is associated with a respective user and comprises a description of a respective predictive model;

selecting a model implementation from a plurality of model implementations for each of the obtained representations based on a status of the representation'"'"'s respective user, the model implementation comprising one or more computer programs operable to execute on one or more computing devices; and

associating each of one or more of the model implementations with a node in a directed graph wherein for one or more ordered pairs of nodes in the graph the prediction output of a model implementation associated with a first node in the pair serves as input to a model implementation associated with a second node in the pair.

2. The method of claim 1, further comprising executing each associated model implementation in an order prescribed by the graph.

3. The method of claim 1 wherein a particular representation is a Predictive Model Markup Language document.

4. The method of claim 1 wherein a particular representation is a textual representation.

5. The method of claim 1 wherein a particular predictive model is a Naive Bayes classifier, a k-nearest neighbor classifier, a support vector machine, or a predictive model that uses a logistic regression technique.

6. The method of claim 5 wherein at least two of the predictive models are different.

7. The method of claim 1 a particular model implementation is a map-reduce implementation.

8. The method of claim 1 a number of computing devices for a particular model implementation is based at least in part on the status of the representation'"'"'s respective user.

9. The method of claim 1 wherein obtaining the plurality of representations further comprises performing pre-processing or post-processing operations in parallel for a particular representation in the plurality of representations.

10. A system comprising:

data processing apparatus configured to execute instructions which cause the data processing apparatus to perform operations comprising;

obtaining a plurality of representations of predictive models, each representation having been received from a different client wherein each representation is associated with a respective user and comprises a description of a respective predictive model;

selecting a model implementation from a plurality of model implementations for each of the obtained representations based on a status of the representation'"'"'s respective user, the model implementation comprising one or more computer programs operable to execute on one or more computing devices; and

associating each of one or more of the model implementations with a node in a directed graph wherein for one or more ordered pairs of nodes in the graph the prediction output of a model implementation associated with a first node in the pair serves as input to a model implementation associated with a second node in the pair.

11. The system of claim 10, further comprising executing each associated model implementation in an order prescribed by the graph.

12. The system of claim 10 wherein a particular model representation is a Predictive Model Markup Language document.

13. The system of claim 10 wherein a particular model representation is a textual representation.

14. The system of claim 10 wherein a particular predictive model is a Naive Bayes classifier, a k-nearest neighbor classifier, a support vector machine, or a predictive model that uses a logistic regression technique.

15. The system of claim 14 wherein at least two of the predictive models are different.

16. The system of claim 10 wherein a particular model implementation is a map-reduce implementation.

17. The system of claim 10 wherein a number of computing devices used by a particular model implementation is based at least in part on the status of the representation'"'"'s respective user.

18. The system of claim 10 wherein obtaining the plurality of representations further comprises performing pre-processing or post-processing operations in parallel for a particular representation in the plurality of representations.

19. A storage medium having instructions stored thereon that, when executed by data processing apparatus, cause the data processing apparatus to perform operations comprising:

obtaining a plurality of representations of predictive models, each representation having been received from a different client wherein each representation is associated with a respective user and comprises a description of a respective predictive model;

selecting a model implementation from a plurality of model implementations for each of the obtained representations based on a status of the representation'"'"'s respective user, the model implementation comprising one or more computer programs operable to execute on one or more computing devices; and

associating each of one or more of the model implementations with a node in a directed graph wherein for one or more ordered pairs of nodes in the graph the prediction output of a model implementation associated with a first node in the pair serves as input to a model implementation associated with a second node in the pair.

20. The storage medium of claim 19, further comprising executing each associated model implementation in an order prescribed by the graph.

21. The storage medium of claim 19 wherein a particular model representation is a Predictive Model Markup Language document.

22. The storage medium of claim 19 wherein a particular model representation is a textual representation.

23. The storage medium of claim 19 wherein a particular predictive model is a Naive Bayes classifier, a k-nearest neighbor classifier, a support vector machine, or a predictive model that uses a logistic regression technique.

24. The storage medium of claim 23 wherein at least two of the predictive models are different.

25. The storage medium of claim 19 wherein a particular model implementation is a map-reduce implementation.

26. The storage medium of claim 19 wherein a number of computing devices used by a particular model implementation is based at least in part on the status of the representation'"'"'s respective user.

27. The storage medium of claim 19 wherein obtaining the plurality of representations further comprises performing pre-processing or post-processing operations in parallel for a particular representation in the plurality of representations.