Generalization Analysis of Wavelet Frame based Neural Network for Function Representations using Compactly supported Gaussian Wavelets
نویسنده
چکیده
A new type of Wavelet Neural Network (WNN) has been proposed to enhance the function approximation and generalization performance generating an optimal size network. In the proposed WNN, the nonlinear activation function is a linear combination of wavelets that can be updated during the network training process. As a result the approximate error is significantly decreased. The RBF network based training method for the proposed WNN is implemented. The main advantage of the proposed WN is that it exploits the attractive features of multi scale wavelet decompositions and the capability of conventional neural networks. The results obtained indicate that the proposed WNN has improved learning ability compared to the conventional architectures. Index Terms wavelet neural network (wnn), continuous wavelet transform, and function representation.
منابع مشابه
ar X iv : m at h / 99 12 12 9 v 1 [ m at h . FA ] 1 5 D ec 1 99 9 COMPACTLY SUPPORTED WAVELETS
We study the harmonic analysis of the quadrature mirror filters coming from multiresolution wavelet analysis of compactly supported wavelets. It is known that those of these wavelets that come from third order polyno-mials are parametrized by the circle, and we compute that the corresponding filters generate irreducible mutually disjoint representations of of the Cuntz algebra O 2 except at two...
متن کاملCompactly Supported Multivariate Wavelet Frames Obtained by Convolution
In this paper, we construct pairs of compactly supported dual wavelet frames for general dilations. We obtain smooth wavelets with small supports by convolution. Our approach leads to wavelets that have a high number of vanishing moments. Moreover, the corresponding refinable function is interpolating. The wavelets for the quincunx dilation matrix in our examples additionally have very high sym...
متن کاملEstimation of Reference Evapotranspiration Using Artificial Neural Network Models and the Hybrid Wavelet Neural Network
Estimation of evapotranspiration is essential for planning, designing and managing irrigation and drainage schemes, as well as water resources management. In this research, artificial neural networks, neural network wavelet model, multivariate regression and Hargreaves' empirical method were used to estimate reference evapotranspiration in order to determine the best model in terms of efficienc...
متن کاملCompactly Supported Symmetric C∞ Wavelets with Spectral Approximation Order
In this paper, we obtain symmetric C∞ real-valued tight wavelet frames in L2(R) with compact support and the spectral frame approximation order. Furthermore, we present a family of symmetric compactly supported C∞ orthonormal complex wavelets in L2(R). A complete analysis of nonstationary tight wavelet frames and orthonormal wavelet bases in L2(R) is given.
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2009