Classification support using confidence intervals
Wilbert van Norden, F. Bolderheij, Catholijn M. Jonker · International Conference on Information Fusion · 2008
In recent years the discrepancy between the required knowledge and the available knowledge for obtaining the situation awareness aboard Royal Netherlands Navy ships has increased. This paper presents a methodology to automatically classify objects in the mission environment based on user defined mission information in order to close this gap. The cornerstone of this methodology is the Confidence Interval (CI) based on Gaussian distributed measurements. To solve the classification problem effectively we describe how such a CI can be constructed for any number of variables. A mission planner has been implemented where operators can define a mission and where objects in the environment are automatically classified based on those mission parameters.