An adaptive newton-like training algorithm for nonlinear filters which have embedded memory
نویسندگان
چکیده
We discuss herein filtering problems involving the emulation of nonlinear systems which have embedded dynamics and where the truth model consists only of an input and output sequence. It has been shown [1] that steepest descent training algorithms such as BackpropagationThrough-Time (BPTT) are locally H optimal for applications where training inputs vary at each weight update, or training epoch. However, when the same data set is used for several epochs, as is typical for nonlinear system identification, BPTT is suboptimal, and a new adaptive GaussNewton technique, which generates updates closer to the Newton update direction, is preferable. We discuss the application of the technique to IIR and FIR filter architectures for preand post-compensating nonlinear systems. Comparisons to canonical methods, namely BPTT, Kalman Filtering, Gauss-Newton, and Broyden-Fletcher-Goldfarb-Shanno (BFGS) are favorably made.
منابع مشابه
Adaptive Steffensen-like Methods with Memory for Solving Nonlinear Equations with the Highest Possible Efficiency Indices
The primary goal of this work is to introduce two adaptive Steffensen-like methods with memory of the highest efficiency indices. In the existing methods, to improve the convergence order applied to memory concept, the focus has only been on the current and previous iteration. However, it is possible to improve the accelerators. Therefore, we achieve superior convergence orders and obtain as hi...
متن کاملAdaptive filtering with the self-organizing map: A performance comparison
In this paper we provide an in-depth evaluation of the SOM as a feasible tool for nonlinear adaptive filtering. A comprehensive survey of existing SOM-based and related architectures for learning input-output mappings is carried out and the application of these architectures to nonlinear adaptive filtering is formulated. Then, we introduce two simple procedures for building RBF-based nonlinear ...
متن کاملUtilizing Kernel Adaptive Filters for Speech Enhancement within the ALE Framework
Performance of the linear models, widely used within the framework of adaptive line enhancement (ALE), deteriorates dramatically in the presence of non-Gaussian noises. On the other hand, adaptive implementation of nonlinear models, e.g. the Volterra filters, suffers from the severe problems of large number of parameters and slow convergence. Nonetheless, kernel methods are emerging solutions t...
متن کاملAdaptive Approximation-Based Control for Uncertain Nonlinear Systems With Unknown Dead-Zone Using Minimal Learning Parameter Algorithm
This paper proposes an adaptive approximation-based controller for uncertain strict-feedback nonlinear systems with unknown dead-zone nonlinearity. Dead-zone constraint is represented as a combination of a linear system with a disturbance-like term. This work invokes neural networks (NNs) as a linear-in-parameter approximator to model uncertain nonlinear functions that appear in virtual and act...
متن کاملAdaptive fuzzy sliding mode and indirect radial-basis-function neural network controller for trajectory tracking control of a car-like robot
The ever-growing use of various vehicles for transportation, on the one hand, and the statistics ofsoaring road accidents resulting from human error, on the other hand, reminds us of the necessity toconduct more extensive research on the design, manufacturing and control of driver-less intelligentvehicles. For the automatic control of an autonomous vehicle, we need its dynamic...
متن کامل