Considerations for the Design of Data Fusion Systems.
Katherine M. Simonson · 2021
In the ATR paradigm, the non-target class consists of everything in the world that is not the designated target.While ample data may be available, it can be difficult to adequately model such a disperse class.-(Note that the ATR paradigm can be readily adapted to multiple target classes.)-In the surveillance paradigm, the goal is to identify deviations from "normal" spatio-temporal patterns.Training data representing off-normal conditions may be limited.