Modeling States of an Entity

• US 20110022369A1
• Filed: 07/27/2009
• Published: 01/27/2011
• Est. Priority Date: 07/27/2009
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Carroll, Melissa Kristin, Cecchi, Guillermo Alberto, Rish, Irina
• Granted Patent: US 10,223,632 B2
• Time in Patent Office: Days
• US Class Current: 703/11
• CPC Class Codes: G06N 3/02 Neural networks

Abstract

Methods, systems and apparatus for modeling states of an entity are presented. For example, a method, implemented on a processor device, of modeling one or more states of an entity is presented. The method includes obtaining a training dataset for training a model by applying a stimulus to the entity, forming a set of model parameters, and using the set of model parameters to form the model, such that the model is configured to predict at least one of the one or more states of the entity. At least one model parameter of the set of model parameters changes with time as a result of dependency of the at least one model parameter on the stimulus and as a result of time-dependency of the stimulus. The steps of obtaining the training dataset, forming the set of model parameters and using the set of model parameters are implemented on the processor device.

Claims

1. A method, implemented on a processor device, of modeling one or more states of an entity, the method comprising the steps of:

obtaining a training dataset for training a model by applying a stimulus to the entity;

forming a set of model parameters, wherein at least one model parameter of the set of model parameters changes with time as a result of dependency of the at least one model parameter on the stimulus and as a result of time-dependency of the stimulus; and

using the set of model parameters to form the model, such that the model is configured to predict at least one of the one or more states of the entity, wherein the steps of obtaining the training dataset, forming the set of model parameters, and using the set of model parameters are implemented on the processor device.

2. The method of claim 1, wherein the model comprises a regression, wherein coefficients of the regression are dependent upon the stimulus.

3. The method of claim 2, wherein the coefficients of the regression are expressed as a Taylor expansion of the stimulus, and wherein the coefficients of the regression are parameters of the set of parameters.

4. The method of claim 3, wherein the parameters are indicated by β

ij(y) and expressed as;

β

ij(y)=α

_0ij+α

_1ijy+α

_2ijy²+

wherein y represents the stimulus; and

wherein α

_0ij, α

_1ij, and α

_2ij are coefficients of the Taylor expansion and i and j are indexes of the parameters.

5. The method of claim 1, wherein the training dataset comprises one or more samples that are non-independent and identically distributed samples (non-i.i.d. samples), and wherein the one or more samples form a time-series involving one or more independent variables.

6. The method of claim 1, wherein the step of forming the set of model parameters comprises application of at least one of:

(i) a Lasso method, (ii) an elastic network method, (iii) a least angle regression selection-elastic network method, and (iv) a least angle regression selection method.

7. The method of claim 1, wherein one or more states of the entity are changing over time.

8. The method of claim 5, wherein the training dataset comprises at least one of:

(i) functional magnetic resonance imaging (fMRI) data and (ii) a time-series.

9. The method of claim 1 adapted to at least one of:

(i) predict brain activity, (ii) predict a mental state, (iii) predict a disease, (iv) discriminate between mental states, and (v) predict brain activity in response to a stimulus.

10. The method of claim 1, wherein the entity comprises a brain, and wherein the one or more states comprise at least one of states brain states and brain activity.

11. The method of claim 1, wherein the model is adapted to indicate a brain area relevant to predicting future brain states.

12. The method of claim 1 further comprising the steps of:

applying the model to the training dataset to produce an accuracy indicator for verifying the model;

comparing the accuracy indicator with a predetermined accuracy threshold; and

increasing a complexity of the model by increasing a complexity of a dependence of an autoregressive parameter on the stimulus when the accuracy indicator is below the predetermined accuracy threshold.

13. The method of claim 1 adapted to perform the steps of:

obtaining a test dataset by applying the stimulus to a test entity; and

predicting a mental state by applying the model to the test dataset and the stimulus.

14. The method of claim 12, wherein one or more of the steps of:

(i) forming the set of model parameters, (ii) using the set of model parameters to form the model, (iii) applying the model to the training dataset, (iv) comparing the accuracy indicator with the predetermined accuracy threshold, and (v) increasing a complexity of the model are repeated until the accuracy indicator is at least one of equal to the predetermined accuracy threshold and above the predetermined accuracy threshold.

15. The method of claim 13, wherein the step of predicting a mental state by applying the model comprises application of at least one of:

(i) a Lasso method, (ii) an elastic network method, (iii) a least angle regression selection-elastic network method, and (iv) a least angle regression selection method.

16. The method of claim 13, wherein the test dataset comprises at least one of:

(i) functional magnetic resonance imaging (fMRI) data and (ii) a time-series.

17. The method of claim 13, wherein the test entity comprises a brain.

18. The method of claim 13, wherein the entity is also the test entity.

19. The method of claim 12, wherein the model comprises a regression having regression terms, and wherein the step of increasing a complexity of the model comprises adding at least one additional regression term.

20. The method of claim 19, wherein coefficients of the regression are expressed as a Taylor expansion of the stimulus, and wherein adding the at least one additional regression term comprises adding at least one higher order term to the Taylor expansion.

21. A system, implemented on a processor device, for modeling one or more states of an entity, the system comprising:

a training data procurement module configured to obtain a training dataset for training a model by applying a stimulus to the entity;

a parameter forming module configured to form a set of model parameters, wherein at least one model parameter of the set of model parameters changes with time as a result of dependency of the at least one model parameter on the stimulus and as a result of time-dependency of the stimulus; and

a model forming module configured to use the set of model parameters to form the model, such that the model is configured to predict at least one of the one or more states of the entity, wherein the training data procurement module, the parameter forming module, and the model forming module are implemented on the processor device.

22. The system of claim 21 further comprising:

a training application module configured to apply the model to the training dataset to produce an accuracy indicator for verifying the model;

a comparison module configured to compare the accuracy indicator with a predetermined accuracy threshold; and

an accuracy module configured to increase a complexity of the model by increasing a complexity of a dependence of an autoregressive parameter on the stimulus when the accuracy indicator is below the predetermined accuracy threshold.

23. The system of claim 21 further comprising:

a test data procurement module configured to obtain a test dataset by applying the stimulus to the test entity; and

a prediction module configured to predict a mental state by applying the model to the test dataset and the stimulus.

24. Apparatus for modeling one or more states of an entity, the apparatus comprising:

a memory; and

a processor coupled to the memory, operative to;

obtain a training dataset for training a model by applying a stimulus to the entity;

form, a set of model parameters, wherein at least one model parameter of the set of model parameters changes with time as a result of dependency of the at least one model parameter on the stimulus and as a result of time-dependency of the stimulus; and

use the set of model parameters to form the model, such that the model is configured to predict at least one of the one or more states of the entity.

25. A computer program product for modeling one or more states of an entity, the computer program product comprising a computer readable storage medium having computer readable program code embodied therewith, the computer readable program code comprising computer readable program code configured to perform the step of:

obtaining a training dataset for training a model by applying a stimulus to the entity;

forming, a set of model parameters, wherein at least one model parameter of the set of model parameters changes with time as a result of dependency of the at least one model parameter on the stimulus and as a result of time-dependency of the stimulus; and

using the set of model parameters to form the model, such that the model is configured to predict at least one of the one or more states of the entity.