Deep symbolic learning and semantics for an explainable and ethical artificial intelligence

Abstract

The main objective of this research is to investigate new hybrid neuro-symbolic algorithms for the construction of an open-source Deep Symbolic Learning framework that allows the training and application of explainable and ethical Deep Learning models. This framework will be supported by an ontology and a layer model in which it is taken into account which user is responsible for interpreting each of the output results according to his or her role, considering, also, the ethical implications of those results.