Neuro Symbolic Reasoning with Ontological Networks

In this chapter we describe neuro symbolic approaches developed for ontological domains, focusing mainly on Recurrent Reasoning Networks (RRNs), a formalism that was recently developed with the goal of automated learning of deep neural networks capable of deciding ontological entailment. After describing the intuition and main details of RRNs, we conclude by briefly mentioning other related approaches, such as Probabilistic Soft Logic (PSL, a formalism based on statistical relational learning) and its recent extension called NeuPSL.