Higher-order statistics symmetry properties

• Tipo: Artigo em Livro de Atas de Conferência Nacional
• Data: 1995

Autores

André Puga (Autor) FEUP; Artur Pimenta Alves (Autor) FEUP

Ata de Conferência Nacional

Título: Proceedings of the 7th Portuguese Conference on Pattern Recognition

Páginas: 311-318

7th Portuguese Conference on Pattern Recognition

Aveiro, 23 - MARÇO - 1995

Outras Informações

Resumo (PT): Higher-Order Statistics are a powerful tool to analyse and classify signals, particularly effective in the presence of non linearities or non gaussianities. The introduction of such methods in engineering and in pattern recognition problems requires the introduction of efficient ways to estimate the higher-order statistical parameters. It has been shown that, due to the symmetry properties of the higher order statistics, the cumulants and moments of 3rd and 4th order, bi- and tri-spectra can be significantly simplified. This paper introduces the properties of symmetry of these parameters and their implications in the simplification of the corresponding estimation. Special attention will be given to tri-spectra aiming to clarify and complement the present literature.

Idioma: Português

Tipo (Avaliação Docente): Científica

Nº de páginas: 8