Inference of Sequential Machines from Sample Computations

This correspondence presents and justifies an algorithm for finding a minimal, though not a smallest, finite sequential machine with a behavior comprising a given finite sample of input-output behavior of some unknown sequential machine. When the number of states n of the machine to be identified is known and the sample contains all input-output pairs of length 2n - 1, the machine found will be equivalent to the original machine and will be minimal.