Automatic determination of report granularity

• US 6,115,702 A
• Filed: 12/23/1997
• Issued: 09/05/2000
• Est. Priority Date: 12/23/1997
• Status: Expired due to Term

• Current Assignee: Hughes Electronics Corporation (AT&T, Inc.)
• Original Assignee: Hughes Electronics Corporation (AT&T, Inc.)
• Inventors: Reiser, Kurt, Chen, Yang
• Primary Examiner(s): Teska, Kevin J.
• Assistant Examiner(s): DAVIS, GEORGE B
• Application Number:
  US08/998,402
• Time in Patent Office: 987 Days
• Field of Search: 706/62, 702/38, 382/103, 382/141
• US Class Current: 706/62
• CPC Class Codes: G06F 18/256 of results relating to different input data, e.g. multimodal recognition

Abstract

The present invention provides a method of performing statistical classification that can resolve conflict in independent sources of information, thereby creating a robust statistical classifier that has superior performance to classifiers currently available. Additionally, the present invention is automatically trainable, yielding improved classification performance. The present invention may be embodied in a method of statistically classifying events or objects, including the steps of gathering sets of information representative of features of an object or event; creating basic probability assignments based on said sets of information; determining a coarse information set from said sets of information; performing coarsening on said sets of information; performing linear superposition on each feature; and combining all features to reach a conclusion.

Claims

1. A method for statistical classification in target recognition, comprising the steps of:

gathering sets of information representative of features of a target;

creating basic probability assignments based on said sets of information;

determining a coarse information set from elements in a choice set;

performing coarsening on said sets of information;

performing linear superposition on each information source; and

combining information representative of said features to reach a conclusion regarding the target.

2. The method of claim 1 wherein said sets of information comprise sensor outputs.

3. The method of claim 1 wherein said sets of information comprise rules.

4. The method of claim 1 wherein said coarse information set is defined by selecting said elements in said choice set having the most conflict.

5. The method of claim 4 wherein said choice set having said most conflict is determined through the use of a conflict operation.

6. The method of claim 4 wherein said choice set has said highest correlation.

7. The method of claim 1 wherein said step of linear superposition includes making a combination of fine, coarse, and vacuous basic probability assignments, each of said basic probability assignments having a coefficient.

8. The method of claim 7 wherein said coefficients may be manipulated to improve classification performance.

9. A method for statistical classification to resolve conflicting sets of information about an object, comprising the steps of:

gathering sets of information representative of features of an object;

creating basic probability assignments based on sail sets of information;

determining a coarse information set from elements in a choice set;

performing coarsening on said sets of information;

performing linear superposition on each information source; and

combining information representative of said features to reach a conclusion regarding the object.