Time series prediction using wavelet and neural networks
Kok Keong Teo · DR-NTU (Nanyang Technological University) · 2002
In this dissertation, the effectiveness of Wavelet Packet Multi-Layer Perceptrons (WP-MLP) neural network for temporal sequence prediction is investigated. The WP-MLP is essentially a neural network with an input signal decomposed to various resolutions using wavelet transform. The Wavelet packets are used to decompose one-dimensional signals so as to illuminate two important properties.