A Preliminary Auditory Subsystem Based on a Growing Functional Modules Controller
In the present paper, a learning based controller that performs as an auditory subsystem is described. Based on the Growing Functional Modules (GFM) paradigm, the auditory subsystem is the result of the interconnection of four kinds of components: Global Goals, Acting Modules, Sensations and Sensing Modules. The resulting controller is radically different from conventional speech recognition due to its ability to gradually learn in context a set of vocal commands while performing in a virtual environment. Recognition is considered as satisfactory when the robot’s behavior is in accordance with the commands’ meaning. This learning process may be compared to how a dog learns to obey its master’s orders. The experiment described in this paper illustrates the design process of a GFM controller. It exhibits the behavior of the control process and principally, exposes the inherent philosophy of the GFM approach.
KeywordsMachine learning knowledge acquisition learning based control connectionism artificial brain
Unable to display preview. Download preview PDF.
- 1.Leboeuf-Pasquier, J., González Pacheco Oceguera, E.: Designing a Growing Functional Modules “Artificial Brain”. Broad Research in Artificial Intelligence and Neuroscience 3(2), 5–15 (2012)Google Scholar
- 3.Leboeuf-Pasquier, J.: A Growing Functional Module Designed to Trigger Causal Inference. In: Proceedings of the 4th Intl Conference on Informatics in Control, Automation & Robotics, ICINCO, pp. 456–463 (2007)Google Scholar
- 4.Galvan, R.: The Equilibrium Subsystem of a Humanoid Robot based on the GFM Paradigm. Master Thesis in Information Technology, CUCEA, University of Guadalajara (2012)Google Scholar
- 5.Gómez-Avila, G.: An Auditory Subsystem based on the GFM Paradigm. Master Thesis in Information Technology, CUCEA, University of Guadalajara (2011)Google Scholar
- 6.Oppenheim, A.V., Schafer, R.W.: From Frequency to Quefrency: A History of the Cepstrum. IEEE Signal Processing Magazine (September 2004)Google Scholar
- 8.Antoniol, G., Rollo, F., Venturi, G.: Linear predictive coding and cepstrum coefficients for mining time variant information from software repositories. In: MSR International Workshop on Mining Software Repositories (2005)Google Scholar