RESUMO
This paper presents a vector quantization process that can be applied online to a stream of inputs. It enables to set up and maintain a dynamical representation of the current information in the stream as a topology preserving graph of prototypical values, as well as a velocity field. The algorithm relies on the formulation of the accuracy of the quantization process, that allows for both the updating of the number of prototypes according to the stream evolution and the stabilization of the representation from which velocities can be extracted. A video processing application is presented.
Assuntos
Algoritmos , Redes Neurais de Computação , Sistemas On-LineRESUMO
In this paper (An abbreviated version of some portions of this article appeared in reference Menard and Frezza-Buet (Menard, O., & Frezza-Buet, H. (2004). Rewarded multi-modal neuronal self-organization: Example of the arm reaching movement. In: Proceedings of international conference on advances in intelligent systems theory and application.), as part of the IJCNN 2005 conference proceedings, published under the IEEE copyright.), an original self-organizing model is presented, with experiments highlighting its ability to be used in different frameworks, as phonetic coding dependent on semantics and arm-reaching. The model relies on the coupling of the learning processes that stand at different self-organizing modules, and exhibits dynamics that can be discussed in terms of the binding of different modalities, scattered over the different modules. Such a binding property is based on an emerging constraint of keeping consistency between the modules. This process is induced by partial connectivity and appropriate neural field competition mechanisms.