Neural nets and finite automata

Finite automata and regular languages were introduced by S.C. Kleene in 1956 to analyze

McCulloch-Pitts nerve nets. In their Turing Prize winning article, Rabin and Scott simplified

finite automata, moving them away from the details of nerve nets in favour of a black box

perspective that made no assumptions about the states of an automaton. This project revisits

Kleene’s treatment of nerve nets, and updates it to take into account some of the many

advances in neural nets and finite automata since then. Results will be reported in a Python

notebook.