A Possible World-Based Fusion Estimation Model for Uncertain Data Clustering in WBNs.

Area of Science:

• Data Science
• Machine Learning
• Probability Theory

Background:

• Data clustering often treats measured data as uncertain.
• Probability-based clustering, like possible worlds, can handle uncertain data.
• Existing possible world methods require determining data and probabilities for each world.

Purpose of the Study:

• To propose a possible world-based fusion estimation model for uncertain data clustering.
• To convert deterministic measurements into probability distributions for natural probability assignment.
• To introduce Kullback-Leibler divergence for analyzing probability distribution relationships.

Main Methods:

• Developed a fusion estimation model to transform deterministic data into probability distributions.
• Utilized Kullback-Leibler divergence to quantify relationships between probability distributions across possible worlds.
• Applied the model to data clustering within wearable body networks (WBNs).

Main Results:

• The proposed model effectively clusters uncertain data by estimating probability distributions.
• Kullback-Leibler divergence provides a method to describe inter-world probability distribution relationships.
• Simulations demonstrate superior performance with complex feature relationships in measured data.

Conclusions:

• The fusion estimation model offers a robust approach to clustering uncertain data using probability distributions.
• The method is particularly effective in applications like wearable body networks.
• The model's performance advantage increases with data complexity.