Modified gradient algorithm for total least square filtering

Kong, Xiangyu; Han, Chongzhao; Wei, Ruixuan

doi:10.1016/j.neucom.2005.10.005

Cited by 17 publications

(14 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [5] and [3], the algorithm(3) is analyzed in detail and it is concluded that the above algorithm is the best TLS neuron in terms of stability (no finite time divergence), speed, and accuracy. In [18], a modified gradient algorithm was proposed, and the algorithm has the best accuracy and the best convergence performance when a larger learning factor is used or SNR is lower in linear system. However, from the analysis [18], the weight norm of above two neurons can be obtained as follows:…”

Section: The Tls Linear Neuron With a Self-stabilizing Algorithmmentioning

confidence: 99%

“…In [18], a modified gradient algorithm was proposed, and the algorithm has the best accuracy and the best convergence performance when a larger learning factor is used or SNR is lower in linear system. However, from the analysis [18], the weight norm of above two neurons can be obtained as follows:…”

Section: The Tls Linear Neuron With a Self-stabilizing Algorithmmentioning

confidence: 99%

“…From (4), we can see that the squared modulus of the augmented weight vector in MCA EXIN algorithm always increases with the oscillatory behavior, and the squared modulus of the augmented weight vector in [18] is divergent when the initial value W(k) 2 is larger than one. Commonly in the TLS problem, the initial value of the squared modulus is bigger than or equal to one, and then W(k) 2 > 1 is true.…”

Section: The Tls Linear Neuron With a Self-stabilizing Algorithmmentioning

confidence: 99%

“…Following the analysis method [5,18], the weight norm of above three algorithms can be obtained as follows:…”

Section: The Tls Linear Neuron With a Self-stabilizing Algorithmmentioning

confidence: 99%

See 3 more Smart Citations

A Self-Stabilizing Neural Algorithm for Total Least Squares Filtering

Kong

Han

2009

Neural Process Lett

View full text Add to dashboard Cite

A neural approach for solving the total least square (TLS) problem is presented in the paper. It is based on a linear neuron with a self-stabilizing neural algorithm, capable of resolving the TLS problem present in the parameter estimation of an adaptive FIR filters for system identification, where noisy errors affect not only the observation vector but also the data matrix. The learning rule is analyzed mathematically and the condition to guarantee the stability of algorithm is educed. The computer simulations are given to illustrate that the neural approach is self-stabilizing and considerably outperforms the existing TLS methods when a larger learning factor is used or the signal-noise-rate is lower.

show abstract

Section: The Tls Linear Neuron With a Self-stabilizing Algorithmmentioning

confidence: 99%

Section: The Tls Linear Neuron With a Self-stabilizing Algorithmmentioning

confidence: 99%

Section: The Tls Linear Neuron With a Self-stabilizing Algorithmmentioning

confidence: 99%

“…Following the analysis method [5,18], the weight norm of above three algorithms can be obtained as follows:…”

Section: The Tls Linear Neuron With a Self-stabilizing Algorithmmentioning

confidence: 99%

See 2 more Smart Citations

A Self-Stabilizing Neural Algorithm for Total Least Squares Filtering

Kong

Han

2009

Neural Process Lett

View full text Add to dashboard Cite

show abstract

“…Least squares method (least squares) is a classical method to solve this problem [1]. A basic assumption of the classical least squares method is that the noise is limited to the output data of the system.…”

Section: Introductionmentioning

confidence: 99%

Research on total least squares method based on error entropy criterion

Wang¹,

Kong²,

Huang³

et al. 2017

Proceedings of the 2nd International Conference on Mechatronics Engineering and Information Technology (ICMEIT 2017)

Self Cite

View full text Add to dashboard Cite

Abstract. The total least square (TLS) estimation problem of random systems is widely found in many fields of engineering and science, such as signal processing, automatic control, system theory and so on. In the case of linear Gaussian case, a very mature TLS parameter estimation algorithm has been developed. In the non-Gaussian case, the existing research is not much and not deep, the current error entropy criterion (EEC) and the minimum error entropy( MEE) based on EEC method has been paid attention to, the traditional MEE only consider the output data contains the error situation, so it cannot get the optimal solution. In this paper, we consider the inclusion of noise in the input and output data, and deduce the total error entropy criterion (TEEC) and the corresponding TLS method named the minimum total error entropy (MTEE) method The In addition, the derivation method is simulated, and the simulation results show the correctness of the algorithm.

show abstract