Fine-tuning enhanced probabilistic neural networks using metaheuristic-driven optimization

• Autor(a) principal: Fernandes, S. E.N.
• Data de Publicação: 2016
• Outros Autores: Setoue, K. K.F. [UNESP], Adeli, H., Papa, J. P. [UNESP]
• Tipo de documento: Capítulo de livro
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.1016/B978-0-12-804536-7.00002-8; http://hdl.handle.net/11449/220833
• topic: Enhanced probabilistic neural networks; Metaheuristic; Neural networks; Optimization; Pattern recognition

Many approaches using neural networks have been studied in the past years. A number of architectures for different objectives are presented in the literature, including probabilistic neural networks (PNNs), which have shown good results in several applications. A simple and elegant solution related to PNNs is the enhanced probabilistic neural networks (EPNNs), whose idea is to consider only the samples that fall in a neighborhood of given a training sample to estimate its probability density function. In this work, we propose to fine-tune EPNN parameters by means of metaheuristic-driven optimization techniques, from the results evaluated in a number of public datasets.