We discuss in this paper architectures for executing probabilistic rule-bases in a paralleludmanner, using as a theoretical basis recently introduced information-theoreticudmodels. We will begin by describing our (non-neural) learning algorithm and theoryudof quantitative rule modelling, followed by a discussion on the exact nature of twoudparticular models. Finally we work through an example of our approach, going fromuddatabase to rules to inference network, and compare the network's performance withudthe theoretical limits for specific problems.
展开▼
