System and Method for Case-Based Multilabel Classification and Ranking

• US 20080010226A1
• Filed: 06/19/2007
• Published: 01/10/2008
• Est. Priority Date: 06/29/2006
• Status: Active Grant

• Current Assignee: Siemens Corp. (Siemens AG)
• Original Assignee: Siemens Corporate Research Incorporated (Siemens AG)
• Inventors: Neubauer, Claus, Brinker, Klaus
• Granted Patent: US 7,860,818 B2
• Time in Patent Office: Days
• US Class Current: 706/012
• CPC Class Codes: G05B 23/0229 knowledge based, e.g. expert systems; genetic algorithms
• CPC Class Codes: G06F 18/2433 Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
• CPC Class Codes: G06N 20/00 Machine learning

Abstract

The present invention provides methods and apparatus for determining and utilizing case-based ranking methods, such as methods for machine condition monitoring. Specifically, the present invention provides a method for identifying and prioritizing labeled data. The method allows a monitored system to be associated with a calibrated and ordered set of states. Further, in machine condition monitoring, the machine condition is associated with the entire set of states in a particular order with one or more relevance zero-points That is, a ranked set of calibrated data describing machine conditions is augmented with an annotation indicating a cut-off between relevant and non-relevant data.

Claims

1. A case-based method of multilabel ranking comprising:

receiving a set of labels partitioned into a first subset of labels and a second subset of labels;

inserting one or more zero-points between the first subset of labels and the second subset of labels;

determining generalized ranks for the labels and the one or more zero-points within the first subset of labels and the second subset of labels;

aggregating the generalized ranks; and

determining a consensus ranking for the labels based on the aggregated generalized ranks for the labels.

2. The method of claim 1 wherein the first subset of labels are relevant and the second subset of labels are non-relevant.

3. The method of claim 1 wherein determining generalized ranks for the labels within the first subset of labels and the second subset of labels comprises:

determining an average overall position of a label within the subset of labels that contains the label.

4. The method of claim 3 wherein the average overall position σ

(i) of a label λ

_i is

5. The method of claim 1 wherein aggregating the generalized ranks comprises:

aggregating the generalized ranks using Spearman'"'"'s footrule metric.

6. The method of claim 1 wherein aggregating the generalized ranks comprises:

aggregating the generalized ranks using Kendall'"'"'s tau.

7. The method of claim 1 wherein determining a consensus ranking for the labels comprises:

determining L(σ

)=min_τ

L(τ

) wherein;

L(σ

) is the distance to a consensus ranking; and

τ

is the label'"'"'s rank correlation coefficient.

8. An apparatus for case-based multilabel ranking comprising:

means for receiving a set of labels partitioned into a first subset of labels and a second subset of labels;

means for inserting one or more zero-points between the first subset of labels and the second subset of labels;

means for determining generalized ranks for the labels within the first subset of labels and the second subset of labels;

means for aggregating the generalized ranks; and

means for determining a consensus ranking for the labels based on the aggregated generalized ranks for the labels.

9. The apparatus of claim 8 wherein the first subset of labels are relevant and the second subset of labels are non-relevant.

10. The apparatus of claim 8 wherein the means for determining generalized ranks for the labels within the first subset of labels and the second subset of labels comprises:

means for determining an average overall position of a label within the subset of labels that contains the label.

11. The apparatus of claim 10 wherein the average overall position σ

(i) of a label λ

_i is

12. The apparatus of claim 8 wherein the means for aggregating the generalized ranks comprises:

means for aggregating the generalized ranks using Spearman'"'"'s footrule metric.

13. The apparatus of claim 8 wherein the means for aggregating the generalized ranks comprises:

means for aggregating the generalized ranks using Kendall'"'"'s tau.

14. The apparatus of claim 8 wherein the means for determining a consensus ranking for the labels comprises:

means for determining L(σ

)=min_τ

L(τ

) wherein;

L(σ

) is the distance to a consensus ranking; and

τ

is the label'"'"'s rank correlation coefficient.

15. A machine readable medium having program instructions stored thereon, the instructions capable of execution by a processor and defining the steps of:

receiving a set of labels partitioned into a first subset of labels and a second subset of labels;

inserting one or more zero-points between the first subset of labels and the second subset of labels;

determining generalized ranks for the labels within the first subset of labels and the second subset of labels;

aggregating the generalized ranks; and

determining a consensus ranking for the labels based on the aggregated generalized ranks for the labels.

16. The machine readable medium of claim 15 wherein the first subset of labels are relevant and the second subset of labels are non-relevant.

17. The machine readable medium of claim 15, wherein the instructions for determining generalized ranks for the labels within the first subset of labels and the second subset of labels further define the step of:

determining an average overall position of a label within the subset of labels that contains the label.

19. The machine readable medium of claim 15, wherein the instructions for aggregating the generalized ranks further define the step of:

aggregating the generalized ranks using Spearman'"'"'s footrule metric.

20. The machine readable medium of claim 15, wherein the instructions for aggregating the generalized ranks further define the step of:

aggregating the generalized ranks using Kendall'"'"'s tau.

21. The machine readable medium of claim 15, wherein the instructions for determining a consensus ranking for the labels further define the step of:

determining L(σ

)=min_τ

L(τ

) wherein;

L(σ

) is the distance to a consensus ranking; and

T is the label'"'"'s rank correlation coefficient.

18. The machine readable medium of claim 18 wherein the average overall position σ

(i) of a label λ

_i is

22. A method of machine condition monitoring comprising:

monitoring a plurality of sensors adapted to detect machine condition information;

evaluating the machine condition information by employing a case-based multilabel algorithm predicting a zero-point between a first subset of labels and a second subset of labels; and

,

outputting an evaluated machine condition from the case-based multilabel algorithm with a zero-point between the first subset of labels and the second subset of labels based on the machine condition information.

23. The method of claim 22 wherein the first subset of labels are relevant and the second subset of labels are non-relevant.

24. The method of claim 22 wherein one of the subsets of labels is an empty set.

25. The method of claim 22 wherein employing a case-based multilabel algorithm comprises:

receiving a set of training data partitioned into a first subset and a second subset;

inserting one or more zero-points between the first subset and the second subset;

determining generalized ranks for the training data and the one or more zero-points within the first subset and the second subset;

aggregating the generalized ranks; and

determining a consensus ranking for the training data based on the aggregated generalized ranks.