Characterizing language identification in terms of computable numberings (Q676311)

Inductive inference is concerned with machines that infer (or learn) the index for a computable function (or r.e. set) from information about that function. The usual model involves getting function values and making conjectures about what the function might be. The conjectures may change over time. There are two very different threads: learning recursive functions from \(f(0),f(1),\ldots\), and learning r.e. sets from just seeing elements that are in the set. Freivalds characterized exactly what classes of functions could be learned in terms of computable numberings. This paper does a similar characterization for learning classes of r.e. sets.