Multivariate Shannon's entropy for adaptive IIR filtering via kernel density estimators

Autor(es)

Fantinato, D. G. ^[1] ; Silva, D. G. ^[2] ; Attux, R. ^[3] ; Neves, A. ^[1]

Afiliação do(s) autor(es)

^[1] Fed Univ ABC, Santo Andre, SP - Brazil

^[2] Univ Brasilia, Brasilia, DF - Brazil

^[3] Univ Estadual Campinas, Campinas, SP - Brazil

• Tipo de documento: Artigo Científico
• Fonte: ELECTRONICS LETTERS; v. 55, n. 15, p. 859+, JUL 25 2019.
• Citações Web of Science: 0

Resumo

In supervised infinite impulse response adaptive filtering, approximate gradient-based approaches are the usual option among optimisation methods. When based on the mean squared error (MSE) criterion, however, these approaches may present biased solutions in noisy scenarios. In that sense, instead of the MSE, the authors propose the use of Shannon's error entropy, an information theoretic learning criterion, which is able to extract higher order statistics from the underlying signals. In particular, a multivariate entropy definition is considered, which is applied to derive a Recursive Prediction Error-based algorithm. The performance analyses are carried out in the context of the supervised channel equalisation problem, with results very favourable to the proposal, in high and low noise level environments. (AU)

• Processo FAPESP: 13/14185-2 - Novas Metodologias de Aprendizado Baseado na Teoria da Informação para Equalização Adaptativa
• Beneficiário: Denis Gustavo Fantinato
• Modalidade de apoio: Bolsas no Brasil - Doutorado
• Processo FAPESP: 17/11488-5 - Análise Multivariada da Estrutura Temporal de Dados para Separação Cega de Fontes no Contexto de Misturas Não Lineares e de Múltiplos Conjuntos de Dados
• Beneficiário: Denis Gustavo Fantinato
• Modalidade de apoio: Bolsas no Brasil - Pós-Doutorado