Making neural nets simple enough to succeed at universal relational generalizationKenneth Kurtz · Binghamton University17 Nov