Achieving "Fairness" in Data Fusion Performance Evaluation Development
Abstract
The key issue for an evaluation (T&E) organization is how to affordably achieve fairness in the application of its PE systems. Our PE framework provides a methodology to accomplish this; viz., the DNN Data Fusion & Resource Management (DF&RM) framework provides the hierarchical PE components for PE solution space and a methodology for mapping PE solution space into various PE problem spaces. The scope of this fairness study for performance evaluation of data fusion (DF) systems is to define a philosophy of fairness that is defendable as a basis for developing a PE system. Sample PE system MoEs need to be defined to understand the PE problem space, PE solution space and the PE problem-to-solution space mapping (i.e., the `rules' to map the alternative PE system design solutions to the needed "Fair" PE capability). Implicitly, we are seeking design guidelines for a "best" PE that balances affordability with fairness as defined above. The first technical contribution of this report is in the reusable and extendable PE solution framework within which all applications-layer approaches to PE known to the authors can be expressed. As such, this PE framework exposes PE system design alternatives to the PE system developer and provides a common framework within which alternative PE systems can be compared.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 30, 2006
- Accession Number
- ADA455903
Entities
People
- Christopher N. Bowman
- James Llinas
- Kedar Sambhoos
- Satyaki G. Dastidar
Organizations
- Calspan-University of Buffalo Research Center