TY - GEN
T1 - Automatic shrinkage tuning based on a system-mismatch estimate for sparsity-aware adaptive filtering
AU - Yamagishi, Masao
AU - Yukawa, Masahiro
AU - Yamada, Isao
N1 - Funding Information:
This work was supported in part by JSPS Grants-in-Aid (26730128).
PY - 2017/6/16
Y1 - 2017/6/16
N2 - Exploiting the sparsity in learning algorithms is a key to achieve excellent performances of adaptive filters. This can be realized by the adaptive proximal forward-backward splitting with carefully chosen parameters. In this paper, we propose an automatic parameter tuning based on a minimization principle of a stochastic approximation of the system-mismatch. The proposed approximation has a Tikhonov-type regularization term, which aims to minimize the disturbance by the update of the adaptive filter and mitigates overfitting to an instantaneous observation. Thanks to these properties, the proposed method realizes adaptive parameter tuning without any user-defined parameters, unlike our previous method that utilizes the user-defined parameter to avoid over-fitting. A numerical example demonstrates the efficacy of the proposed parameter tuning.
AB - Exploiting the sparsity in learning algorithms is a key to achieve excellent performances of adaptive filters. This can be realized by the adaptive proximal forward-backward splitting with carefully chosen parameters. In this paper, we propose an automatic parameter tuning based on a minimization principle of a stochastic approximation of the system-mismatch. The proposed approximation has a Tikhonov-type regularization term, which aims to minimize the disturbance by the update of the adaptive filter and mitigates overfitting to an instantaneous observation. Thanks to these properties, the proposed method realizes adaptive parameter tuning without any user-defined parameters, unlike our previous method that utilizes the user-defined parameter to avoid over-fitting. A numerical example demonstrates the efficacy of the proposed parameter tuning.
KW - Sparsity-aware adaptive filter
KW - adaptive proximal forward-backward splitting algorithm
KW - automatic parameter tuning
UR - http://www.scopus.com/inward/record.url?scp=85023748433&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85023748433&partnerID=8YFLogxK
U2 - 10.1109/ICASSP.2017.7953068
DO - 10.1109/ICASSP.2017.7953068
M3 - Conference contribution
AN - SCOPUS:85023748433
T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
SP - 4800
EP - 4804
BT - 2017 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2017 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2017 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2017
Y2 - 5 March 2017 through 9 March 2017
ER -