Realization of common statistical methods in computational linguistics with functional automata

In this paper we present the functional automata as a general framework for representation, training and exploring of various statistical models as LLM’s, HMM’s, CRF’s, etc. Our contribution is a new construction that allows the representation of the derivatives of a function given by a functional automaton. It preserves the natural representation of the functions and the standard product and sum operations of real numbers. In the same time it requires no additional overhead for the standard dynamic programming techniques that yield the computation of a functional value.