Implementation and Use of Algorithms for Conversion and Learning in a Neuro-Symbolic System
Neuro-Symbolic Integration, KBANN, Connectionism, Artificial Neural Networks.
One of the main goals of artificial intelligence is the creation of agents with human-like intelligence. This has been researched using various approaches, and among the most prominent for machine learning are logic-based symbolic systems and artificial neural networks. Until the last decade, both approaches have progressed independently, but progress in both areas has led researchers to investigate ways to integrate both approaches. Several models that provide hybrid or integrated integration of these approaches emerged in the 1990s, and continue to be used to this day. This work has as main objective the implementation of the Neuro-Symbolic conversion algorithm of the KBANN (Knowledge-Based Artificial Neural Networks) hybrid system, which has the ability to map the dependencies of a specific domain of rules (if-then) in a Neural network, and then refine that network using learning techniques. In addition, since this model does not have the ability to refine the network topology in order to obtain new rules for the initial domain, it was necessary to implement another algorithm to the original KBANN model, in order to obtain possible expansions of the original network. KBANN.