Method and apparatus for resource allocation among classifiers in classification systems

1. A method for optimizing resource allocation among data analysis functions in a classification system comprising the steps of:

characterizing each of the data analysis functions as a set of operating points in multi-dimensional space corresponding to a plurality of dimensions of resource requirements and analysis quality;

selecting an operating point for each of the data analysis functions in accordance with one or more constraints; and

applying the analysis functions at selected operating points to optimize resource allocation among the data analysis functions in the classification system,

wherein one or more of the steps of characterizing, selecting and applying are performed by a computer.

2. The method of claim 1 , wherein the classification system comprises a large scale cascaded classification system.

3. The method of claim 1 , wherein, in the step of selecting an operating point, the one or more constraints comprise at least one of a classification system constraint and a user constraint.

4. The method of claim 1 , wherein, in the step of characterizing each of the data analysis functions, each operating point corresponds to a classification algorithm.

5. The method of claim 1 , wherein the classification system comprises a plurality of classification stages, and each of the classification stages comprises one or more data analysis functions.

6. The method of claim 5 , wherein the plurality of classification stages comprise at least one of feature extraction, atomic modeling and composite modeling.

7. The method of claim 1 , wherein, in the step of characterizing each of the data analysis functions, the set of operating points are represented in a multidimensional space corresponding to one or more dimensions of resource requirements and analysis quality.

8. The method of claim 1 , wherein, in the step of characterizing each of the data analysis functions, the resource requirements comprise at least one of computational, storage, communications and human requirements.

9. The method of claim 1 , wherein, in the step of characterizing each of the data analysis functions, the analysis quality comprises at least one of average precision, classification rate, model validity assessments, receiver operating characteristic curves, reject curves, and precision curves.

10. The method of claim 1 , wherein, in the step of characterizing each of the data analysis functions, each data analysis function is characterized offline as meta data.

11. The method of claim 1 , wherein optimization occurs at each classification stage of the classification system.

12. The method of claim 1 , wherein optimization occurs at a final stage of the classification system and incorporate variability of prior classification stages.

13. The method of claim 1 , wherein optimization occurs at an individual classification stage of the classification system when constraints are capable of being allocated to the individual classification stage and optimization occurs at a final stage of the classification system when constraints are not capable of being allocated to prior classification stages.

14. The method of claim 1 , wherein, in the step of characterizing each of the data analysis functions, the set operating points result from an application of at least one of feature extraction, clustering, classification and visualization methods.

15. The method of claim 1 , wherein the step of selecting an operating point comprises the step of defining a constrained optimization problem.

16. The method of claim 15 , wherein the constrained optimization problem maximizes analysis quality with respect to at least one of available resources and response time.

17. The method of claim 15 , wherein the constrained optimization problem minimizes resource requirements with respect to a target analysis quality.

18. The method of claim 15 , wherein the constrained optimization problem is converted into an unconstrained optimization problem in which operating points are selected given the one or more constraints using a Lagrangian method.

19. Apparatus for optimizing resource allocation among data analysis functions in a classification system, comprising:

a memory; and

at least one processor coupled to the memory and operative to: (i) characterize each of the data analysis functions as a set of operating points in multi-dimensional space corresponding to a plurality of dimensions of resource requirements and analysis quality; (ii) select an operating point for each of the data analysis functions in accordance with one or more constraints; and (iii) apply the analysis functions at selected operating points to optimize resource allocation among the data analysis functions in the classification system.

20. An article of manufacture for optimizing resource allocation among data analysis functions in a classification system, comprising a machine readable medium containing one or more programs which when executed implement the steps of:

characterizing each of the data analysis functions as a set of operating points in multi-dimensional space corresponding to a plurality of dimensions of resource requirements and analysis quality;

selecting an operating point for each of the data analysis functions in accordance with one or more constraints; and

applying the analysis functions at selected operating points to optimize resource allocation among the data analysis functions in the classification system.