Evaluation of the Effectiveness of Machine-based Situation Assessment - Preliminary Work

摘要

The Information Fusion Panel within The Technical Cooperation Program (TTCP) is developing algorithms to perform machine-based situation assessment to assist human operators in complex situations. This report proposes a technique to measure the effectiveness of these algorithms in a simulation environment where ground truth is well-defined. In addition, this report models the situation assessment algorithms abstractly using random inference networks, and examines how errors (damage) spread through the inference networks. This models deficiencies in the object assessment as input to the situation assessment algorithm.