Bayesian probabilistic reasoning in design

The application of Bayesian reasoning to engineering design is discussed. The decisions involved in the design process can be represented as nodes in a Bayesian network. The conditional probability that a child node in this network will take a desired value, given the values of its parents, can be estimated from a database built up from simulations of past designs, or from suitably encoded empirical principles. Bayesian probabilistic reasoning can then be used to calculate the probability of particular choices leading to a successful design. In this way the most promising alternatives can be investigated first