Probabilistic automaton classifier applied to examples related to the Andrews-Curtis conjecture

The unsolved Andrews-Curtis conjecture in group theory gives rise to many famously challenging theorem-proving problems. Automated reasoning is one of the best state-of-the-art approaches to solving them. We have developed a new classifier inspired by group theory and a new search algorithm orientated at using for the Andrews-Curtis conjecture. We combine them with automated reasoning and compare their performance with automated reasoning.