Título: Comparison of Parkinson’s Disease Diagnosis using SOM and MLP Neural Networks
Autores: Torres, Camilo A.; Mora, Diego A.; Orjuela-Canon, A. D.
Resumo: This paper realizes a comparison of two methods used as solution to the classification problem of patient’s with Parkinson’s Disease from measures taken on bosses of voice. The used methods are Multilayer Perceptron (MLP) and Kohonen’s self-organizing maps (SOM). Both methods are addressed with and without data preprocessing using Principal Components Analysis (PCA). The best classification result obtained is with MLP without data preprocessing, which achieved a high correct average classification rate of 90.24%, therefore it can be taken into account for the Parkinson’s Disease diagnosis. Finally a comparison between the addressed solutions is made in terms of correct classification rate.
Palavras-chave: Diagnosis; Multilayer Perceptrons; Parkinson ́s Disease; PCA; Self Organizing Maps; supervised learning; unsupervised learning
Código DOI: 10.21528/CBIC2013-343
Artigo em pdf: bricsccicbic2013_submission_343.pdf
Arquivo BibTex: bricsccicbic2013_submission_343.bib