Skip to Main Content
Certain information-processing limitations in hypothesis testing can be modeled as quantization of prior probabilities. While quantization hurts performance, a team of decision makers can minimize their performance loss by adopting diverse quantizers and collaborating on the design of their decision rules. In this paper, the benefits of diversity and collaboration in binary hypothesis testing are discussed. A set of N diverse K-level quantizers used by a team of N collaborating decision makers is as powerful as a single (N(K - 1) + 1)-level quantizer used by them all. If the decision makers do not collaborate, a set of diverse quantizers is less powerful, but it is still better than a set of identical quantizers.