Abstract
The Dempster-Shafer (DS) theory is a powerful general framework for reasoning under uncertainty. While the strength of the DS theoretic (DST) framework in its ability to handle a wider variety of data imperfections is not in dispute, a major criticism cast towards DST evidential reasoning is the heavy computational burden it entails. If the advantages offered by DS theory is to be fully realized, it is essential that one explores efficient data structures and algorithms that can be used for DST operations and computations. In this paper, we wish to present a novel generalized computational framework for exactly this purpose. We develop three representations - DS-Vector, DS-Matrix, and DS-Tree - which allow DST computation to be performed in significantly less time. These three representations can also be utilized as tools for visualizing DST models. A new strategy, which we refer to as REGAP, which allows REcursive Generation of and Access to Propositions is introduced and harnessed in the development of this framework and computational algorithms. The paper also provides a discussion and experimental validation of the utility, efficiency, and implementation of the proposed data structures and algorithms.
| Original language | English (US) |
|---|---|
| Title of host publication | FUSION 2016 - 19th International Conference on Information Fusion, Proceedings |
| Publisher | Institute of Electrical and Electronics Engineers Inc. |
| Pages | 1570-1577 |
| Number of pages | 8 |
| ISBN (Electronic) | 9780996452748 |
| State | Published - Aug 1 2016 |
| Event | 19th International Conference on Information Fusion, FUSION 2016 - Heidelberg, Germany Duration: Jul 5 2016 → Jul 8 2016 |
Other
| Other | 19th International Conference on Information Fusion, FUSION 2016 |
|---|---|
| Country | Germany |
| City | Heidelberg |
| Period | 7/5/16 → 7/8/16 |
Fingerprint
Keywords
- algorithms
- belief functions
- computational frameworks
- data structures
- Dempster-Shafer theory
- evidential reasoning
ASJC Scopus subject areas
- Statistics, Probability and Uncertainty
- Computer Science Applications
- Computer Vision and Pattern Recognition
- Signal Processing
Cite this
A Framework for efficient computation of belief theoretic operations. / Polpitiya, Lalintha G.; Premaratne, Kamal; Murthi, Manohar; Sarkar, Dilip.
FUSION 2016 - 19th International Conference on Information Fusion, Proceedings. Institute of Electrical and Electronics Engineers Inc., 2016. p. 1570-1577 7528070.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - A Framework for efficient computation of belief theoretic operations
AU - Polpitiya, Lalintha G.
AU - Premaratne, Kamal
AU - Murthi, Manohar
AU - Sarkar, Dilip
PY - 2016/8/1
Y1 - 2016/8/1
N2 - The Dempster-Shafer (DS) theory is a powerful general framework for reasoning under uncertainty. While the strength of the DS theoretic (DST) framework in its ability to handle a wider variety of data imperfections is not in dispute, a major criticism cast towards DST evidential reasoning is the heavy computational burden it entails. If the advantages offered by DS theory is to be fully realized, it is essential that one explores efficient data structures and algorithms that can be used for DST operations and computations. In this paper, we wish to present a novel generalized computational framework for exactly this purpose. We develop three representations - DS-Vector, DS-Matrix, and DS-Tree - which allow DST computation to be performed in significantly less time. These three representations can also be utilized as tools for visualizing DST models. A new strategy, which we refer to as REGAP, which allows REcursive Generation of and Access to Propositions is introduced and harnessed in the development of this framework and computational algorithms. The paper also provides a discussion and experimental validation of the utility, efficiency, and implementation of the proposed data structures and algorithms.
AB - The Dempster-Shafer (DS) theory is a powerful general framework for reasoning under uncertainty. While the strength of the DS theoretic (DST) framework in its ability to handle a wider variety of data imperfections is not in dispute, a major criticism cast towards DST evidential reasoning is the heavy computational burden it entails. If the advantages offered by DS theory is to be fully realized, it is essential that one explores efficient data structures and algorithms that can be used for DST operations and computations. In this paper, we wish to present a novel generalized computational framework for exactly this purpose. We develop three representations - DS-Vector, DS-Matrix, and DS-Tree - which allow DST computation to be performed in significantly less time. These three representations can also be utilized as tools for visualizing DST models. A new strategy, which we refer to as REGAP, which allows REcursive Generation of and Access to Propositions is introduced and harnessed in the development of this framework and computational algorithms. The paper also provides a discussion and experimental validation of the utility, efficiency, and implementation of the proposed data structures and algorithms.
KW - algorithms
KW - belief functions
KW - computational frameworks
KW - data structures
KW - Dempster-Shafer theory
KW - evidential reasoning
UR - http://www.scopus.com/inward/record.url?scp=84992129668&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84992129668&partnerID=8YFLogxK
M3 - Conference contribution
SP - 1570
EP - 1577
BT - FUSION 2016 - 19th International Conference on Information Fusion, Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
ER -