Adaptive sensor fusion with nets of binary threshold elements

A simple distributed-detection scheme whose probability of error can be calculated analytically is demonstrated, and it is shown that it corresponds to a two-layer network of binary threshold elements. The authors assume that the sensors and the fusion center are subject to sudden unpredictable changes in the environment that they survey and show how learning algorithms can be used in order to maintain good performance, in spite of these changes. They conclude with an example involving five unequal sensors which distinguish between two time-varying Gaussian populations of different means.<>