> Because EMX is a new model in machine learning, we do not yet know its usefulness for developing real-world algorithms. So these results might not turn out to

> Because EMX is a new model in machine learning, we do not yet know its usefulness for developing real-world algorithms. So these results might not turn out to have practical importance. I don't think that's the correct conclusion. More likely, all learning models have unprovable problems so the choice of model is not important wrt the existence of unprovable problems. Really, this is not a surprising result. All mathematical systems gave unprovable problems. ML is a mathematical system and no exception.