Nonlinear time series forecasting with Bayesian neural networks

A novel hybrid learning algorithm for full Bayesian approach of artificial neural networks

“…Weight decay and Laplace function approaches are well-known for function [7,28]. Essentially, the weight decay is given as a natural interpretation in the Gaussian approximation of Bayesian learning introduced by Mackay [2].…”

“…In spite of incorporating many complicated concepts in its structure, Genetic MC can be implemented to the nonlinear regression and time series analyses easily, and then it allows estimating the robust models in context of training BNNs. Essentially, this novel approach is based on the hybrid learning procedures proposed by Kocadaglı [27], Kocadaglı and Asikgil [28] in which GAs is integrated with MCMC methods for Gaussian approach of BNNs. However, in this study, Genetic MC is adapted to the full Bayesian approach instead of Gaussian approach.…”

“…Kocadaglı [27] integrated the hierarchical Bayesian learning with GAs and the fuzzy numbers, and then he proposed Genetic MC algorithm in the context of full Bayesian approach. Kocadaglı and Aş ıkgil [28] adapted Genetic MC, proposed by Kocadaglı [27], to Gaussian approximation of BNNs.…”

A unified model for context-based behavioural modelling and classification

“…Wang et al (2015) propose the use of an adaptive differential evolution algorithm to select appropriate initial connection weights and thresholds. Kocadaǧlı and Aşıkgil (2014) use a Bayesian inference approach to train a ANN. An evolutionary Monte Carlo algorithm is proposed.…”

Previsões Pontuais E Intervalares De Séries Temporais De Alta Frequência Com Sistema De Lógica Fuzzy

“…Estes podem ser classificados em, basicamente, quatro categorias: modelos estatísticos (que podem ser subdivididos em modelos de estatística clássica [1,3] e de estatística bayesiana [4,5]), modelos de inteligência computacional (que podem ser subdivididos em modelos de lógica fuzzy [6,7], de redes neurais [8,9], de support vector regression [10,11] e outros), modelos híbridos (que combinam diferentes modelos, e.g. [12,13]), e outros tipos de modelos (e.g. judgemental forecasting [14]).…”