9th European Signal Processing Conference (EUSIPCO 1998), 1998
The binormalized data-reusing least mean squares (BNDR-LMS) algorithm has been recently proposed ... more The binormalized data-reusing least mean squares (BNDR-LMS) algorithm has been recently proposed and has been shown to have faster convergence than other LMS-like algorithms in cases where the input signal is strongly correlated. This superior performance in convergence speed is, however, followed by a higher misadjustment if the step-size is close to the value which allows the fastest convergence. An optimal step-size sequence for this algorithm is proposed after considering a number of simplifying assumptions. Moreover, this work brings insight in how to deal with these conflicting requirements of fast convergence and minimum steady-state mean square error (MSE).
Page 1. ON FAST QR ALGORITHMS BASED ON BACKWARD PREDICTION ERRORS: NEW RESULTS AND COMPARISONS Jo... more Page 1. ON FAST QR ALGORITHMS BASED ON BACKWARD PREDICTION ERRORS: NEW RESULTS AND COMPARISONS José A. Apolinário Jr. 1 , Marcio G. Siqueira 2 , and Paulo SR Diniz 3 ½Escuela Politécnica del Ejército Av. ...
ABSTRACT This paper applies data selective updating to the Modified Conjugate Gradient algorithm.... more ABSTRACT This paper applies data selective updating to the Modified Conjugate Gradient algorithm. In search for a new conjugate-gradient-like filtering algorithm, two different approaches are developed: the first one results in the recently proposed set-membership affine projection (SM-AP) algorithm and the second one reduces the computational requirements of the modified congujate gradient algorithm while keeping approximately the same good results in terms of conver-gence speed and misadjustment. Simulation results for a system identification experiment show the claimed perfor-mance with a considerable reduced number of updates.
This paper presents the performance of a text independent speaker verification system using Gauss... more This paper presents the performance of a text independent speaker verification system using Gaussian Mixture Model (GMM) for the Brazilian Portuguese. The Gaussian compo-nents of the GMM statistically represent the spectral char-acteristics of the speaker, leading to an effective speaker recognition system. The main goal here is a detailed evalua-tion of the parameters used by the GMM such as the number of Gaussian mixtures, the amount of time for training and testing. Aiming at the definition of the best set of features for a reasonable response, this work helps the comprehen-sion of the model and gives insights for further investiga-tion. We have used 36 speakers in the experiments, all mod-eled with 15 mel-cepstral coefficients. For 32 Gaussians, 60 seconds of training, and 30 seconds of testing, the sys-tem has no failure for a reasonably clean speech signal. The results have shown that the higher the amount of time for training and testing, the better are the results for a give...
2001 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.01CH37221), 2001
ABSTRACT Adaptive filtering techniques in subbands have been recently developed for a number of a... more ABSTRACT Adaptive filtering techniques in subbands have been recently developed for a number of applications including acoustic echo cancellation and wideband active noise control. In such applications, hundreds of taps are required resulting in high computational complexity and low convergence rate when using LMS-based algorithms. For fullband systems, new algorithms which try to overcome these drawbacks have been investigated. A class of these algorithms employing variants of the filtered gradient adaptive (FGA) algorithm has been successfully developed. We apply these techniques to a recently proposed subband adaptive filter structure in order to improve the convergence rate and the computational load. Computer simulations show the benefits obtained with these proposed algorithms
2013 IEEE 4th Latin American Symposium on Circuits and Systems (LASCAS), 2013
ABSTRACT This paper investigates the performance of a fast converging adaptive filter, the Recurs... more ABSTRACT This paper investigates the performance of a fast converging adaptive filter, the Recursive Least Squares algorithm based on the Inverse QR Decomposition (IQRD-RLS), with an exact initialization procedure, for the online estimation of lowdamped electromechanical modes in a power system. In this approach, the modes are tracked from ambient data, once it is assumed that load variations constantly excite the electromechanical dynamics as a nearly white noise input. Monte Carlo linear simulations are run on the full Brazilian Interconnected Power System model to generate power system ambient data. The performance of the IQRD-RLS algorithm is compared to that of the Least Mean Squares (LMS) algorithm when estimating the slowest interarea mode in the system.
9th European Signal Processing Conference (EUSIPCO 1998), 1998
The binormalized data-reusing least mean squares (BNDR-LMS) algorithm has been recently proposed ... more The binormalized data-reusing least mean squares (BNDR-LMS) algorithm has been recently proposed and has been shown to have faster convergence than other LMS-like algorithms in cases where the input signal is strongly correlated. This superior performance in convergence speed is, however, followed by a higher misadjustment if the step-size is close to the value which allows the fastest convergence. An optimal step-size sequence for this algorithm is proposed after considering a number of simplifying assumptions. Moreover, this work brings insight in how to deal with these conflicting requirements of fast convergence and minimum steady-state mean square error (MSE).
Page 1. ON FAST QR ALGORITHMS BASED ON BACKWARD PREDICTION ERRORS: NEW RESULTS AND COMPARISONS Jo... more Page 1. ON FAST QR ALGORITHMS BASED ON BACKWARD PREDICTION ERRORS: NEW RESULTS AND COMPARISONS José A. Apolinário Jr. 1 , Marcio G. Siqueira 2 , and Paulo SR Diniz 3 ½Escuela Politécnica del Ejército Av. ...
ABSTRACT This paper applies data selective updating to the Modified Conjugate Gradient algorithm.... more ABSTRACT This paper applies data selective updating to the Modified Conjugate Gradient algorithm. In search for a new conjugate-gradient-like filtering algorithm, two different approaches are developed: the first one results in the recently proposed set-membership affine projection (SM-AP) algorithm and the second one reduces the computational requirements of the modified congujate gradient algorithm while keeping approximately the same good results in terms of conver-gence speed and misadjustment. Simulation results for a system identification experiment show the claimed perfor-mance with a considerable reduced number of updates.
This paper presents the performance of a text independent speaker verification system using Gauss... more This paper presents the performance of a text independent speaker verification system using Gaussian Mixture Model (GMM) for the Brazilian Portuguese. The Gaussian compo-nents of the GMM statistically represent the spectral char-acteristics of the speaker, leading to an effective speaker recognition system. The main goal here is a detailed evalua-tion of the parameters used by the GMM such as the number of Gaussian mixtures, the amount of time for training and testing. Aiming at the definition of the best set of features for a reasonable response, this work helps the comprehen-sion of the model and gives insights for further investiga-tion. We have used 36 speakers in the experiments, all mod-eled with 15 mel-cepstral coefficients. For 32 Gaussians, 60 seconds of training, and 30 seconds of testing, the sys-tem has no failure for a reasonably clean speech signal. The results have shown that the higher the amount of time for training and testing, the better are the results for a give...
2001 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.01CH37221), 2001
ABSTRACT Adaptive filtering techniques in subbands have been recently developed for a number of a... more ABSTRACT Adaptive filtering techniques in subbands have been recently developed for a number of applications including acoustic echo cancellation and wideband active noise control. In such applications, hundreds of taps are required resulting in high computational complexity and low convergence rate when using LMS-based algorithms. For fullband systems, new algorithms which try to overcome these drawbacks have been investigated. A class of these algorithms employing variants of the filtered gradient adaptive (FGA) algorithm has been successfully developed. We apply these techniques to a recently proposed subband adaptive filter structure in order to improve the convergence rate and the computational load. Computer simulations show the benefits obtained with these proposed algorithms
2013 IEEE 4th Latin American Symposium on Circuits and Systems (LASCAS), 2013
ABSTRACT This paper investigates the performance of a fast converging adaptive filter, the Recurs... more ABSTRACT This paper investigates the performance of a fast converging adaptive filter, the Recursive Least Squares algorithm based on the Inverse QR Decomposition (IQRD-RLS), with an exact initialization procedure, for the online estimation of lowdamped electromechanical modes in a power system. In this approach, the modes are tracked from ambient data, once it is assumed that load variations constantly excite the electromechanical dynamics as a nearly white noise input. Monte Carlo linear simulations are run on the full Brazilian Interconnected Power System model to generate power system ambient data. The performance of the IQRD-RLS algorithm is compared to that of the Least Mean Squares (LMS) algorithm when estimating the slowest interarea mode in the system.
Uploads
Papers by Jose Antonio Apolinario Jr.