In a binary communication system, the symbols 0 and 1 are sent through a channel, but noise can switch 1 to 0 or vice versa with certain probabilities. The error can be estimated by Bayes' rule.

Let be the event of sending 0 (represented by ) and be the event of sending 1 (represented by ), with corresponding probabilities and . On the receiving side, and are the events of receiving 0 or 1. Represented by and , is the probability of receiving 1 given that 0 was sent, is the probability of receiving 1 given that 1 was sent, is the conditional probability of given , and so on.