Tracking methods and systems that employ bayesian networks

1. A device for tracking a real-time behavior of a system, the device comprising:

a processor; and

a memory connected to the processor configured to store a dynamic relational Bayesian network (DRBN);

wherein the processor is configured to perform:

identifying variables in a DRBN to be filtered, the DRBN comprising a plurality of submodels;

creating a reduced graph for each submodel of the DRBN by reducing each observed variable to a singleton domain;

compiling each reduced graph to create compiled inference structures;

creating a mapping between data instances and the compiled inference structures by constructing an inference cache for each variable of the reduced graph for each data instance to which the reduced graph applies; and

performing exact inference over each data instance using a generalized loopy inference method to communicate estimate changes between instances for each loopy pass.

2. The device of claim 1 , wherein the processor is further configured to perform assigning a characterization to the variables to be filtered, and wherein the characterization is a Boolean characterization.

3. The device of claim 1 , wherein the DRBN comprises a plurality of classes of submodels, each class comprising a plurality of submodels, and wherein each of the submodels for each class is associated with a condition of applicability, and wherein at least some nodes of some submodels are labeled as import and update; and

wherein the processor is further configured to perform:

classifying each of the submodels as static, dynamic or dependent;

following loopy inference over each new instance, for all inference node caches where the instance is of a submodel labeled dynamic and has a node labeled import and update, locating the instance containing the variable corresponding to the node labeled import;

repeating the locating n times; and

scavenging the instance located in the nth locating.

4. The device of claim 1 , wherein the instances are historical data instances and the inference is used for parameter estimation.

5. The device of claim 1 , wherein the instances are real time data instances and the inference is used for filtering.

6. The device of claim 1 , wherein the DRBN is a model of a web visit.

7. The device of claim 1 , wherein the DRBN comprises a page view class of submodels, and wherein the page view class comprises at least one first submodel of a page view of a page showing a specific product and at least one second submodel of a page view of a page not showing a specific product.

8. The device of claim 1 , wherein the system is in-residence patient behavior.

9. The device of claim 1 , wherein the system is a collection of moving, physical targets.

10. The device of claim 1 , wherein the system is a visitor on a website.

11. The device of claim 1 , wherein the data instances are time ordered and the order of instance application is reversed for each loopy pass.

12. A non-transitory computer-readable medium having instructions stored thereon that, when executed by a processor, cause the processor to track real-time behavior of a system, the tracking comprising:

identifying variables in a dynamic relational Bayesian network (DRBN) to be filtered, the DRBN including a plurality of submodels;

creating a reduced graph for each submodel of the DRBN by reducing each observed variable to a singleton domain;

compiling each reduced graph to create compiled inference structures;

creating a mapping between data instances and the compiled inference structures by constructing an inference cache for each variable of the reduced graph for each data instance to which the reduced graph applies; and

performing exact inference over each data instance using a generalized loopy inference method to communicate estimate changes between instances for each loopy pass.

13. The non-transitory computer-readable medium of claim 12 , the tracking further comprising assigning a characterization to the variables to be filtered;

wherein the characterization is a Boolean characterization.

14. The non-transitory computer-readable medium of claim 12 , wherein the DRBN comprises a plurality of classes of submodels, each class comprising a plurality of submodels, and wherein each of the submodels for each class is associated with a condition of applicability, and wherein at least some nodes of some submodels are labeled as import and update; and

wherein the tracking further comprises:

classifying each of the submodels as static, dynamic or dependent;

following loopy inference over each new instance, for all inference node caches where the instance is of a submodel labeled dynamic and has a node labeled import and update, locating the instance containing the variable corresponding to the node labeled import;

repeating the locating n times; and

scavenging the instance located in the nth locating.

15. The non-transitory computer-readable medium of claim 12 , wherein the instances are historical data instances and the inference is used for parameter estimation.

16. The non-transitory computer-readable medium of claim 12 , wherein the instances are real time data instances and the inference is used for filtering.

17. The non-transitory computer-readable medium of claim 12 , wherein the DRBN comprises a page view class of submodels, and wherein the page view class comprises at least one first submodel of a page view of a page showing a specific product and at least one second submodel of a page view of a page not showing a specific product.

18. A computer-implemented method for tracking real-time behavior of a system, the computer-implemented method comprising:

identifying variables in a dynamic relational Bayesian network (DRBN) to be filtered, the DRBN including a plurality of submodels;

creating a reduced graph for each submodel of the DRBN by reducing each observed variable to a singleton domain;

compiling each reduced graph to create compiled inference structures;

creating a mapping between data instances and the compiled inference structures by constructing an inference cache for each variable of the reduced graph for each data instance to which the reduced graph applies; and

performing exact inference over each data instance using a generalized loopy inference method to communicate estimate changes between instances for each loopy pass.

19. The computer-implemented method of claim 18 , wherein the DRBN comprises a plurality of classes of submodels, each class comprising a plurality of submodels, and wherein each of the submodels for each class is associated with a condition of applicability, and wherein at least some nodes of some submodels are labeled as import and update; and

wherein the computer-implemented method further comprises:

classifying each of the submodels as static, dynamic or dependent;

following loopy inference over each new instance, for all inference node caches where the instance is of a submodel labeled dynamic and has a node labeled import and update, locating the instance containing the variable corresponding to the node labeled import;

repeating the locating n times; and

scavenging the instance located in the nth locating.

20. The computer-implemented method of claim 18 , wherein the DRBN comprises a page view class of submodels, and

wherein the page view class comprises at least one first submodel of a page view of a page showing a specific product and at least one second submodel of a page view of a page not showing a specific product.