Maximum partial likelihood framework for channel equalization by distribution learning

Tulay Adali, Xiao Liu, Ning Li, Mustafa (Kemal) Sonmez

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

We present the general formulation for adaptive equalization by distribution learning in which conditional probability mass function (pmf) of the transmitted signal given the received is parametrized by a general neural network structure. The parameters of the pmf are computed by minimization of the accumulated relative entropy (ARE) cost function. The equivalence of ARE minimization to maximum partial log-likelihood (MPLL) estimation is established under certain regularity conditions which enables us to bypass the requirement that the true conditionals be known. The large sample properties of MPLL estimator are obtained under further regularity conditions, and the binary case with sigmoidal perceptron as the conditional pmf model is shown to be a special case of the new framework. Results are presented which show that the multilayer perceptron (MLP) equalizer based on ARE minimization can always recover from convergence at the wrong extreme whereas the mean square error (MSE) based MLP can not.

Original languageEnglish (US)
Title of host publicationNeural Networks for Signal Processing - Proceedings of the IEEE Workshop
PublisherIEEE
Pages541-550
Number of pages10
StatePublished - 1995
Externally publishedYes
EventProceedings of the 5th IEEE Workshop on Neural Networks for Signal Processing (NNSP'95) - Cambridge, MA, USA
Duration: Aug 31 1995Sep 2 1995

Other

OtherProceedings of the 5th IEEE Workshop on Neural Networks for Signal Processing (NNSP'95)
CityCambridge, MA, USA
Period8/31/959/2/95

Fingerprint

Entropy
Multilayer neural networks
Neural networks
Equalizers
Mean square error
Cost functions

ASJC Scopus subject areas

  • Signal Processing
  • Software
  • Electrical and Electronic Engineering

Cite this

Adali, T., Liu, X., Li, N., & Sonmez, M. K. (1995). Maximum partial likelihood framework for channel equalization by distribution learning. In Neural Networks for Signal Processing - Proceedings of the IEEE Workshop (pp. 541-550). IEEE.

Maximum partial likelihood framework for channel equalization by distribution learning. / Adali, Tulay; Liu, Xiao; Li, Ning; Sonmez, Mustafa (Kemal).

Neural Networks for Signal Processing - Proceedings of the IEEE Workshop. IEEE, 1995. p. 541-550.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Adali, T, Liu, X, Li, N & Sonmez, MK 1995, Maximum partial likelihood framework for channel equalization by distribution learning. in Neural Networks for Signal Processing - Proceedings of the IEEE Workshop. IEEE, pp. 541-550, Proceedings of the 5th IEEE Workshop on Neural Networks for Signal Processing (NNSP'95), Cambridge, MA, USA, 8/31/95.
Adali T, Liu X, Li N, Sonmez MK. Maximum partial likelihood framework for channel equalization by distribution learning. In Neural Networks for Signal Processing - Proceedings of the IEEE Workshop. IEEE. 1995. p. 541-550
Adali, Tulay ; Liu, Xiao ; Li, Ning ; Sonmez, Mustafa (Kemal). / Maximum partial likelihood framework for channel equalization by distribution learning. Neural Networks for Signal Processing - Proceedings of the IEEE Workshop. IEEE, 1995. pp. 541-550
@inproceedings{42f8dd4dd7664464a4a2699010419ea6,
title = "Maximum partial likelihood framework for channel equalization by distribution learning",
abstract = "We present the general formulation for adaptive equalization by distribution learning in which conditional probability mass function (pmf) of the transmitted signal given the received is parametrized by a general neural network structure. The parameters of the pmf are computed by minimization of the accumulated relative entropy (ARE) cost function. The equivalence of ARE minimization to maximum partial log-likelihood (MPLL) estimation is established under certain regularity conditions which enables us to bypass the requirement that the true conditionals be known. The large sample properties of MPLL estimator are obtained under further regularity conditions, and the binary case with sigmoidal perceptron as the conditional pmf model is shown to be a special case of the new framework. Results are presented which show that the multilayer perceptron (MLP) equalizer based on ARE minimization can always recover from convergence at the wrong extreme whereas the mean square error (MSE) based MLP can not.",
author = "Tulay Adali and Xiao Liu and Ning Li and Sonmez, {Mustafa (Kemal)}",
year = "1995",
language = "English (US)",
pages = "541--550",
booktitle = "Neural Networks for Signal Processing - Proceedings of the IEEE Workshop",
publisher = "IEEE",

}

TY - GEN

T1 - Maximum partial likelihood framework for channel equalization by distribution learning

AU - Adali, Tulay

AU - Liu, Xiao

AU - Li, Ning

AU - Sonmez, Mustafa (Kemal)

PY - 1995

Y1 - 1995

N2 - We present the general formulation for adaptive equalization by distribution learning in which conditional probability mass function (pmf) of the transmitted signal given the received is parametrized by a general neural network structure. The parameters of the pmf are computed by minimization of the accumulated relative entropy (ARE) cost function. The equivalence of ARE minimization to maximum partial log-likelihood (MPLL) estimation is established under certain regularity conditions which enables us to bypass the requirement that the true conditionals be known. The large sample properties of MPLL estimator are obtained under further regularity conditions, and the binary case with sigmoidal perceptron as the conditional pmf model is shown to be a special case of the new framework. Results are presented which show that the multilayer perceptron (MLP) equalizer based on ARE minimization can always recover from convergence at the wrong extreme whereas the mean square error (MSE) based MLP can not.

AB - We present the general formulation for adaptive equalization by distribution learning in which conditional probability mass function (pmf) of the transmitted signal given the received is parametrized by a general neural network structure. The parameters of the pmf are computed by minimization of the accumulated relative entropy (ARE) cost function. The equivalence of ARE minimization to maximum partial log-likelihood (MPLL) estimation is established under certain regularity conditions which enables us to bypass the requirement that the true conditionals be known. The large sample properties of MPLL estimator are obtained under further regularity conditions, and the binary case with sigmoidal perceptron as the conditional pmf model is shown to be a special case of the new framework. Results are presented which show that the multilayer perceptron (MLP) equalizer based on ARE minimization can always recover from convergence at the wrong extreme whereas the mean square error (MSE) based MLP can not.

UR - http://www.scopus.com/inward/record.url?scp=0029239287&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0029239287&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:0029239287

SP - 541

EP - 550

BT - Neural Networks for Signal Processing - Proceedings of the IEEE Workshop

PB - IEEE

ER -