Stochastic Gradient Algorithms with a Gradient-Adaptive and Limited Step-Size

Akihiko SUGIYAMA

Stochastic Gradient Algorithms with a Gradient-Adaptive and Limited Step-Size

Akihiko SUGIYAMA

Full Text Views

0

Share
Cite this

Summary :

This paper proposes new algorithms for adaptive FIR filters. The proposed algorithms provide both fast convergence and small final misadjustment with an adaptive step size even under an interference to the error. The basic algorithm pays special attention to the interference which contaminates the error. To enhance robustness to the interference, it imposes a special limit on the increment/decrement of the step-size. The limit itself is also varied according to the step-size. The basic algorithm is extended for application to nonstationary signals. Simulation results with white signals show that the final misadjustment is reduced by up to 22 dB under severe observation noise at a negligible expense of the convergence speed. An echo canceler simulation with a real speech signal exhibits its potential for a nonstationary signal.

Publication: IEICE TRANSACTIONS on Fundamentals Vol.E77-A No.3 pp.534-538

Publication Date: 1994/03/25

Publicized

Online ISSN

DOI

Type of Manuscript: PAPER

Category: Adaptive Signal Processing

Cite this

Copy

Akihiko SUGIYAMA, "Stochastic Gradient Algorithms with a Gradient-Adaptive and Limited Step-Size" in IEICE TRANSACTIONS on Fundamentals, vol. E77-A, no. 3, pp. 534-538, March 1994, doi: .
Abstract: This paper proposes new algorithms for adaptive FIR filters. The proposed algorithms provide both fast convergence and small final misadjustment with an adaptive step size even under an interference to the error. The basic algorithm pays special attention to the interference which contaminates the error. To enhance robustness to the interference, it imposes a special limit on the increment/decrement of the step-size. The limit itself is also varied according to the step-size. The basic algorithm is extended for application to nonstationary signals. Simulation results with white signals show that the final misadjustment is reduced by up to 22 dB under severe observation noise at a negligible expense of the convergence speed. An echo canceler simulation with a real speech signal exhibits its potential for a nonstationary signal.
URL: https://globals.ieice.org/en_transactions/fundamentals/10.1587/e77-a_3_534/_p

Copy

@ARTICLE{e77-a_3_534,
author={Akihiko SUGIYAMA, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Stochastic Gradient Algorithms with a Gradient-Adaptive and Limited Step-Size},
year={1994},
volume={E77-A},
number={3},
pages={534-538},
abstract={This paper proposes new algorithms for adaptive FIR filters. The proposed algorithms provide both fast convergence and small final misadjustment with an adaptive step size even under an interference to the error. The basic algorithm pays special attention to the interference which contaminates the error. To enhance robustness to the interference, it imposes a special limit on the increment/decrement of the step-size. The limit itself is also varied according to the step-size. The basic algorithm is extended for application to nonstationary signals. Simulation results with white signals show that the final misadjustment is reduced by up to 22 dB under severe observation noise at a negligible expense of the convergence speed. An echo canceler simulation with a real speech signal exhibits its potential for a nonstationary signal.},
keywords={},
doi={},
ISSN={},
month={March},}

Copy

TY - JOUR
TI - Stochastic Gradient Algorithms with a Gradient-Adaptive and Limited Step-Size
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 534
EP - 538
AU - Akihiko SUGIYAMA
PY - 1994
DO -
JO - IEICE TRANSACTIONS on Fundamentals
SN -
VL - E77-A
IS - 3
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - March 1994
AB - This paper proposes new algorithms for adaptive FIR filters. The proposed algorithms provide both fast convergence and small final misadjustment with an adaptive step size even under an interference to the error. The basic algorithm pays special attention to the interference which contaminates the error. To enhance robustness to the interference, it imposes a special limit on the increment/decrement of the step-size. The limit itself is also varied according to the step-size. The basic algorithm is extended for application to nonstationary signals. Simulation results with white signals show that the final misadjustment is reduced by up to 22 dB under severe observation noise at a negligible expense of the convergence speed. An echo canceler simulation with a real speech signal exhibits its potential for a nonstationary signal.
ER -