Energy minimization for data merging and fusion

• US 6,968,342 B2
• Filed: 06/19/2001
• Issued: 11/22/2005
• Est. Priority Date: 12/29/1997
• Status: Expired due to Term

• Current Assignee: Abel Wolman, Jeff B. Glickman
• Original Assignee: Abel Wolman, Jeff B. Glickman
• Inventors: Wolman, Abel, Glickman, Jeff B.
• Primary Examiner(s): AMSBURY, WAYNE P
• Application Number:
  US09/885,342
• Publication Number:
  US 20020002555A1
• Time in Patent Office: 1,617 Days
• Field of Search: 707/102
• US Class Current: 1/1
• CPC Class Codes: G06F 18/2137 based on criteria of topology preservation, e.g. multidimensional scaling or self-organising maps
• CPC Class Codes: Y10S 707/99943 Generating database or data structure, e.g. via user interface

Abstract

A data merging/fusion comprises using a preprocessing step, an energy minimization step, and a postprocessing step to merge or fuse data. In a particular embodiment, ordinal data are processed by mapping the ordinal data to a lower triangular matrix of ordinal data, processing the matrix using non-metric individual differences multidimensional scaling and subsequently processing the result.

Claims

1. A data fusion method comprising the steps of:

(a) receiving data; and

(b) applying energy minimization to simultaneously produce a plurality of fusion vectors.

2. The data fusion method of claim 1 wherein step (b) comprises steps of:

(b1) creating a plurality of data structures using the received data; and

(b2) simultaneously producing the plurality of fusion vectors from the plurality of data structures.

3. The data fusion method of claim 2 wherein step (b1) comprises compressing the plurality of data structures to produce a new data structure, the new Jata structure having fewer elements than any data structure of the plurality of data structures.

4. The data fusion method of claim 2 wherein the plurality of data structures comprise two or more hollow symmetric matrices.

5. The data fusion method of claim 4 wherein the two or more matrices comprise hollow symmetric matrices filled with values of the received data.

7. The data fusion method of claim 1 wherein the received data is ordinal or higher measurement level data.

8. The data fusion method of claim 1 wherein step (b) comprises applying individual differences multidimensional scaling to the data.

9. The data fusion method of claim 1 further comprising the steps of:

(c) taking norms of the plurality of fusion vectors.

6. A data fusion method comprising the steps of:

(a) receiving data and transposing the received data to produce data structures;

(b) applying energy minimization to the data structures to simultaneously produce a plurality of merged values;

(c) constructing a matrix from the merged values to define a proximity weight matrix; and

(d) applying individual differences multidimensional scaling to data structures derived from the received data and the proximity weight matrix to simultaneously produce a plurality of fusion vectors.

10. A data fusion method comprising the steps of:

(a) receiving ordinal measurement level data;

(b) creating hollow symmetric matrices using the ordinal measurement level data;

(c) applying meaningful energy minimization to the matrices; and

(e) simultaneously producing a plurality of fusion vectors in response to the meaningful energy minimization.

11. The data fusion method of claim 10 wherein step (b) comprises the steps of:

(b

     1) entry-wise populating the hollow symmetric matrices with elements of a collection of sets containing the ordinal measurement level data.

12. The data fusion method of claim 11 further comprising the step of:

(b2) augmenting the data with artificial values.

13. The data fusion method of claim 12 wherein step (b2) comprises:

replacing one or more missing values of the data with artificial values.

14. The data fusion method of claim 12 wherein step (b2) comprises:

appending one or more artificial values to the data to completely fill the hollow symmetric matrices with elements of the collection of sets containing the ordinal measurement level data and the appended artificial values.

15. The data fusion method of claim 12 wherein the artificial values comprise randomly determined values.

16. The data fusion method of claim 12 wherein the artificial values comprise a repeated constant value.

17. The data fusion method of claim 10 wherein the ordinal measurement level data comprises ranking values for a predetermined characteristic among a plurality of input domains.

18. The data fusion method of claim 10 wherein the ordinal measurement level data comprises rating values for a predetermined characteristic among a plurality of input domains.

19. A data fusion process for data, the process comprising:

using individual differences multidimensional scaling with two or more input hollow symmetric matrices into which the data for fusion has been entered to produce at least a source space output; and

using the source space output to interpret the fused data.

20. A data fusion process comprising:

forming input hollow symmetric matrices with ordinal measurement level data;

applying a meaningful energy minimization process with the input hollow symmetric matrices to simultaneously produce at least a plurality of fusion vectors; and

interpreting the plurality of fusion vectors as indicative of data fusion.

21. The data fusion process of claim 20 wherein the forming step comprises the steps of:

receiving the ordinal measurement level data;

treating each ordinal measurement level datum as a proximity;

defining the input hollow symmetric matrices as non-redundant lower triangles of the hollow symmetric matrices of sufficient dimension to contain all the ordinal measurement level data; and

mapping consecutive ordinal measurement level data entry-wise into the non-redundant lower triangles of the hollow symmetric matrices.

22. The data fusion process of claim 21 wherein the ordinal measurement level data comprises ranking values for a predetermined characteristic among a plurality of input domains.

23. The data fusion process of claim 21 wherein the ordinal measurement level data comprises rating values for a predetermined characteristic among a plurality of input domains.

24. The data fusion process of claim 21 wherein the forming step comprises appending one or more artificial values to the data to completely fill the non-redundant lower triangles of the hollow symmetric matrices with the ordinal measurement level data and the artificial values.

25. The data fusion process of claim 20 wherein the meaningful energy minimization process comprises non-metric individual differences multidimensional scaling using the input hollow symmetric matrices, the data fusion process further comprising producing a source space output.

26. The data fusion process of claim 25 further comprising the steps of:

producing source space fusion vectors in the source space output; and

taking a norm of each of the source space fusion vectors.

27. The data fusion process of claim 20 wherein the forming step comprises the steps of:

receiving the data;

replicating the data; and

mapping consecutive replicated data entry-wise into the hollow symmetric matrices.

28. Computer executable software program code stored on a computer readable medium, the code for data fusion of input data, the code comprising:

first code that receives data and forms one or more data structures using the received data;

second code that applies an energy minimization process to the one or more data structures and simultaneously produces a plurality of fusion vectors; and

third code that uses the fusion vectors to provide user output information.

29. A method for data fusion, the method comprising the steps of:

receiving two or more sets of ordinal measurement level data, the ordinal measurement level data including ranking values for a predetermined characteristic among a plurality of input domains;

mapping the ordinal measurement level data into non-redundant lower triangles of two or more hollow symmetric matrices.

processing the hollow symmetric matrices using a meaningful energy minimization process to produce output data; and

processing the output data to provide fused data.

30. A method for data fusion comprising the steps of:

receiving two or more sets of ordinal measurement level data S_k, the ordinal measurement level data including ranking values for a predetermined characteristic among a plurality of input domains;

mapping the ordinal measurement level data into non-redundant lower triangles T_k of two or more hollow symmetric matrices;

processing the matrices T_k using a meaningful energy minimization process to produce output data, including

processing the two or more sets of ordinal data S_k in the matrices T_k as relationally linked deformable configurations to find a configuration corresponding to minimum energy, and

producing the output data associated with the configuration corresponding to minimum energy; and

processing the output data to provide fused data.

31. A data fusion process comprising:

receiving ordinal measurement level data;

transposing the received ordinal measurement level data;

writing the transposed ordinal measurement level data to two or more hollow symmetric matrices;

using energy minimization and the hollow symmetric matrices to produce a source space output;

using the source space output to produce a proximity weight matrix;

writing the received ordinal measurement level into two or more hollow symmetric matrices;

using meaningful energy minimization and the proximity weight matrix and the hollow symmetric matrices to produce a source space output; and

interpreting the source space output as an indicator of data fusion or merging.

32. A method for data merging, the method comprising the steps of:

receiving two or more sets of data;

mapping the received data into two or more hollow symmetric matrices;

processing the resulting hollow symmetric matrices using individual differences multidimensional scaling;

producing a source space of vectors; and

interpreting the source space vectors as indicators of data merging.

33. The data merging method of claim 32 wherein mapping the received data comprises mapping the received data entry-wise into the two or more hollow symmetric matrices.

34. The data merging method of claim 33 wherein mapping the received data comprises augmenting the received data with artificial values so as to fill the hollow symmetric matrices during the entry-wise mapping.

35. The data merging method of claim 33 wherein mapping the received data comprises treating empty entries in the hollow symmetric matrices during entry-wise mapping of the received data as missing values during processing of the hollow symmetric matrices using individual differences multidimensional scaling.

36. The data merging method of claim 32 further comprising the steps of:

varying the dimensionality of the source space; and

associating to each set of received data a set of source space vectors of varying dimensionality.

37. The data merging method of claim 32 further comprising the steps of:

taking a norm of each of the source space vectors; and

interpreting the norms of the source space vectors as merged values of the received sets of data.

39. The data merging method of claim 35 wherein the received ordinal measurement level data is mapped entry-wise into the hollow symmetric matrices.

40. The data merging method of claim 39 wherein the received ordinal measurement level data is augmented with artificial values so as to fill the hollow symmetric matrices under the entry-wise mapping.

41. The data merging method of claim 39 wherein empty entries in the hollow symmetric matrices under entry-wise mapping of the received data are treated as missing values under processing of the hollow symmetric matrices using non-metric individual differences multidimensional scaling.

38. A method for data merging, the method comprising the steps of:

receiving two of more sets of ordinal measurement level data;

mapping the ordinal measurement level data into two or more hollow symmetric matrices;

processing the resulting hollow symmetric matrices using non-metric individual differences multidimensional scaling;

producing a source space of vectors; and

interpreting the source space vectors as indicators of data merging.

42. The data merging method of claim 38 further comprising the steps of:

varying the source space dimensionality; and

associating to each set of received ordinal measurement level data a set comprised of source space vectors of varying dimensionality.

43. The data merging method of claim 38 further comprising the steps of:

taking a norm of each of the source space vectors; and

interpreting the norms of the source space vectors as fused or merged values of the sets of received ordinal measurement level data.

44. The data merging method of claim 43 wherein the source space vectors are interpreted as vectors comprised of ratio measurement level components.

45. The data merging method of claim 43 wherein the norms of the source space vectors are interpreted as ratio measurement level merged values.

46. A data fusion process for data, the process comprising:

using non-metric individual differences multidimensional scaling with two one or more input hollow symmetric matrices into which the data for fusion has been entered to produce at least a source space output; and

using the source space output to interpret the fused data.

47. The data fusion process of claim 46 further comprising the steps of:

prior to the step of using non-metric individual differences multidimensional scaling, receiving ordinal measurement level data; and

producing the two or more input hollow symmetric matrices from the ordinal measurement level data.

48. The data fusion process of claim 47 wherein the producing step comprises treating the ordinal measurement level data as proximity data for input to a non-metric individual differences multidimensional scaling process.