Probabilistic Multilayer Perceptron with Applications to Data Fitting

Nieto-Chaupis H. · International Conference on Engineering and Emerging Technologies, ICEET (2025) · DOI: 10.1109/iceet67911.2025.11423919

Resumen

Among the most important objectives inside of territory of the statistical schemes becomes the prediction of nonlinear system from a little knowledge of past. However, such systems might not be well understood so that would not have space for reliable predictions. In this paper, more than to go through routes that apparently try to model nonlinear systems, it is chosen a fitting mechanism based entirely in the theory of artificial neural networks. Thus for example, when system behaves fully nonlinear, one can employ mathematical methodologies based in perceptron. Under this view, input and weight of perceptron can be adjustable in order to seek the mathematical function more representable or that adjusts well to system. Applications to test the proposed approach of this paper are presented.

Instituciones: Universidad Autónoma del Perú