Sistema neurosimbólico para construção de árvores filogenéticas.

Abstract

This work presents a neurosymbolic system for the construction of phylogenetic trees called SINCA. In this system symbolic and connectionist techniques work cooperatively. The SINCA connectionist module uses a Hopfield neural network to find the shortest distance between the branches of the phylogenetic tree, controlling the combinatorial explosion generated by the number of possible phylogenetic trees for the set of phylogenetic trees. species investigated. The symbolist module of SINCA uses an expert system to build phylogenetic trees from user-provided knowledge of its knowledge base, and the knowledge generated by its connectionist module. A study of phylogenetic trees is presented, the main algorithms for the construction of phylogenetic trees, a study of Hopfield neural networks their stability and their weights, the implementation of a neural algorithm for phylogenetic tree construction (ANCA) and an example of phylogenetic tree construction with ANCA. Finally, we present the implementation of SINCA, some examples of phylogenetic tree construction with the SINCA and suggestions are made for future work.