Abstract

STERPIN BUITRAGO, Dante Giovanni. Imitative cognition for a robot through a community of neuro-memetic replicators (CNMR). Tecnura [online]. 2015, vol.19, n.44, pp.15-31. ISSN 0123-921X.  https://doi.org/10.14483/udistrital.jour.tecnura.2015.2.a01.

Memetic computation, as an artificial intelligence tool, uses models of certain cerebral elements, called neuro-memes, hypothetically implicated in symbolization, dissemination and evolution of cultural characteristics in the human societies. In order to simulate the great evolutionary potential of neuro-memes, the Self-supervised perceptron (SSP) was recently presented as an artificial neuro- meme, and its imitative learning capability was verified through the control of a simple robot. Regarding that capability, it was observed an apprehensive deficiency when the imitator looks only the instructor´s freely evidenced behavior. When attending that deficiency, it was found the incapability to imitate a behavior with certain characteristics. Whereas the SSP is a model oriented to develop collective systems, this paper presents an SSP-based multi-agent system, by which such difficulties found in a single SSP were solved.

Keywords : Artificial neural network; Imitative learning; Kohonen's self-organized map; Multi-agent system; Multilayer perceptron; Neuro-meme; Neuro- memetic replication.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )