Adaptive decision process

• US 20060184482A1
• Filed: 01/10/2006
• Published: 08/17/2006
• Est. Priority Date: 02/14/2005
• Status: Abandoned Application

• Current Assignee: Manyworlds Incorporated
• Original Assignee: Manyworlds Incorporated
• Inventors: Flinn, Steven Dennis, Moneypenny, Naomi Felina
• Application Number:
  US11/329,722
• Publication Number:
  US 20060184482A1
• Time in Patent Office: Days
• US Class Current: 706/45
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks

Abstract

An adaptive decision process is disclosed for more effectively and efficiently determining and conducting information gathering and evaluation associated with decisions. The adaptive decision process integrates decision analysis, value of information analysis, design of experiment models, and the inferencing of gathered information, including experimental results. The process enables an automatic, adaptive, closed-loop process for attaining additional information and assimilating the attained information into the decision model.

Claims

1. A method comprising:

establishing a decision that is influenced by one or more uncertain variables;

identifying one or more actions expected to reduce uncertainty associated with the one or more uncertain variables; and

determining automatically the next one or more actions to conduct based, at least in part, on the automatic evaluation of the results of previously performing one or more actions.

2. The method of claim 1, wherein determining automatically the next one or more actions to conduct based, at least in part, on the automatic evaluation of the results of previously performing one or more actions comprises:

applying an experimental design model.

3. The method of claim 1, wherein determining automatically the next one or more actions to conduct based, at least in part, on the automatic evaluation of the results of previously performing one or more actions comprises:

determining the expected value of information of the one or more actions.

4. The method of claim 3, wherein determining the expected value of information of one or more actions comprises:

determining the net expected value of information of the one or more actions.

5. The method of claim 4, wherein determining the net expected value of information of one or more actions comprises:

incorporating an expected cost associated with the expected time required to attain the results of the one or more actions.

6. The method of claim 3, wherein determining the expected value of information of the one or more actions comprises:

applying a modeling method to determine the expected value of an item of information associated with resolving one or more uncertainties.

7. The method of claim 6, wherein applying a modeling method to determine the expected value of an item of information associated with resolving one or more uncertainties comprises:

applying a modeling method, wherein the modeling method is selected from a group consisting of factorial matrix model, D-optimal design model, regression model, principal component analysis model, Bayesian network model, neural network model, statistical learning model, support vector machine model, decision tree model, decision lattice model, and dynamic programming model.

8. The method of claim 1, wherein determining automatically the next one or more actions to conduct based, based at least in part, on the automatic evaluation of the results of previously performing one or more actions comprises:

applying a statistical model to the information attained as a result of the one or more actions, wherein the statistical model is selected from a group consisting of inductive model, transductive model, regression model, principal component analysis model, statistical learning model, Bayesian model, neural network model, genetic algorithm-based statistical model, and support vector machine model.

9. The method of claim 8, wherein applying a statistical model to the information attained comprises:

integrating the statistical model with a design of experiment model.

10. The method of claim 9, wherein integrating the statistical model with a design of experiment model further comprises:

enabling an adaptive design of experiment process, wherein the adaptive design of experiment process dynamically adjusts an experimental design based, at least in part, on one or more inferences derived from applying the statistical model to the results of previously performing one or more actions.

11. The method of claim 1, wherein determining automatically the next one or more actions to conduct based, at least in part, on the automatic evaluation of the results previously performing one or more actions comprises:

enabling an adaptive design of experiment process, wherein the adaptive design of experiment process dynamically adjusts an experimental design based, at least in part, on one or more inferences derived from applying the statistical model to the results of previously performing one or more actions.

12. The method of claim 11, wherein enabling an adaptive design of experiment process, wherein the adaptive design of experiment process dynamically adjusts an experimental design based, at least in part, on one or more inferences derived from applying the statistical model to the results of previously performing one or more actions comprises:

enabling an automatic feedback means between an information inferencing statistical model and a design of experiment model.

13. The method of claim 12, further comprising:

determining automatically the net expected value of information associated with one or more actions based on one or more inferences from a statistical model and a design of experiment model.

14. The method of claim 11, further comprising:

determining a plurality of sets of actions; and

determining the optimal set of actions.

15. The method of claim 14, further comprising:

determining a plurality of sets of uniquely sequenced actions; and

determining the optimal set of uniquely sequenced actions.

16. The method of claim 1, wherein determining automatically the next one or more actions to conduct based, at least in part, on the automatic evaluation of the results of previously performing one or more actions comprises:

conducting automatically the next one or more actions.

17. The method of claim 16, wherein conducting automatically the next one or more actions comprises:

applying automated information gathering means, wherein the automated information gathering means is selected from a group consisting of computer-based information search, computer-based information retrieval, computer-based human expert network, computer-based data analysis, computer-based process control, computer-based apparatus control, and robotic experimentation apparatus.

18. A method of determining and implementing an information gathering means, the method comprising:

establishing one or more decisions that are influenced by one or more uncertain variables;

identifying one or more simulated information gathering means;

determining the expected one or more actions associated with the one or more decisions to be performed with the one or more simulated information gathering means;

determining the net expected value of the one or more actions;

determining the net expected value of the one or more simulated information gathering means; and

determining the one or more simulated information gathering means that should be implemented.

19. The method of claim 18, wherein determining the expected one or more actions associated with the one or more decisions to be performed with the one or more simulated information gathering means comprises:

applying an automatic value of information function.

20. A system comprising:

a representation of a decision that is influenced by one or more uncertain variables;

a representation of one or more actions expected to reduce uncertainty associated with the one or more uncertain variables; and

means for determining automatically the next one or more actions to conduct based, at least in part, on the automatic evaluation of the results of previously performing one or more actions.

21. The system of claim 20, wherein means for determining automatically the next one or more actions to conduct based on the automatic evaluation of the results of the one or more actions comprises:

an experimental design model.

22. The system of claim 20, wherein means for determining automatically the next one or more actions to conduct based, at least in part, on the automatic evaluation of the results of previously performing one or more actions comprises:

means for determining the net expected value of information of the one or more actions.

23. The system of claim 22, wherein means for determining the net expected value of information of one or more actions comprises:

a model to determine the expected value of an item of information associated with resolving one or more uncertainties.

24. The system of claim 23, wherein a model to determine the expected value of an item of information associated with resolving one or more uncertainties comprises:

a model, wherein the model is selected from a group consisting of factorial matrix model, D-optimal design model, regression model, principal component analysis model, Bayesian network model, neural network model, statistical learning model, support vector machine model, decision tree model, decision lattice model, and dynamic programming model.

25. The system of claim 20, wherein means for determining automatically the next one or more actions to conduct based, at least in part, on the automatic evaluation of the results of previously performing one or more actions comprises:

a statistical model applied to the information attained as a result of the one or more actions, wherein the statistical model is selected from a group consisting of inductive model, transductive model, regression model, principal component analysis model, statistical learning model, Bayesian model, neural network model, genetic algorithm-based statistical model, and support vector machine model.

26. The system of claim 25, wherein a statistical model applied to the information attained comprises:

an integrated statistical model and a design of experiment model.

27. The system of claim 26, wherein an integrated statistical model and a design of experiment model comprises:

an adaptive design of experiment system, wherein the adaptive design of experiment process dynamically adjusts an experimental design based, at least in part, on one or more inferences derived from applying the statistical model to the results of previously performing one or more actions.

28. The system of claim 20, wherein means for determining automatically the next one or more actions to conduct based, at least in part, on the automatic evaluation of the results of previously performing one or more actions comprises:

means to conduct automatically the next one or more actions.

29. The system of claim 28, wherein means to conduct automatically the next one or more actions comprises:

an automated information gathering means, wherein the automated information gathering means is selected from a group consisting of a computer-based information search system, a computer-based information retrieval system, a computer-based data analysis system, computer-based human expert network system, a computer-based process control system, a computer-based apparatus control system, and a robotic experimentation apparatus.