Distributed estimation and joint probabilities estimation by entropy model
2001
This paper proposes the use of Entropy Model for distributed estimation system. Entropy Model is an entropic technique based on the minimization of conditional entropy and developed for Multi-Source/Sensor Information Fusion (MSIF) problem. We address the problem of distributed estimation from independent observations involving multiple sources, i.e., the problem of estimating or selecting one of several identity declaration, or hypothesis concerning an observed object. Two problems are considered in Entropy Model. In order to fuse observations using Entropy Model, it is necessary to know or estimate the conditional probabilities and by equivalent the joint probabilities. A common practice for estimating probability distributions from data when nothing is known (without a priori knowledge), one should prefer distributions that are as uniform as possible, that is, have maximal entropy. Next, the problem of combining (or “fusing”) observations relating to identity hypotheses and selecting the most appropria...
Keywords:
- Information diagram
- Joint entropy
- Maximum-entropy Markov model
- Maximum entropy probability distribution
- Principle of maximum entropy
- Differential entropy
- Maximum entropy thermodynamics
- Artificial intelligence
- Maximum entropy spectral estimation
- Pattern recognition
- Mathematics
- Conditional entropy
- Algorithm
- Entropy estimation
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
9
References
1
Citations
NaN
KQI