Hyper-parameter optimization of iterative control algorithms for adaptive optical systems

Yuxiang LUO; Huizhen YANG; Yuanfeng HE; Zhiguang ZHANG

doi:10.5768/JAO202445.0102007

Journal of Applied Optics, Volume. 45, Issue 1, 126(2024)

Hyper-parameter optimization of iterative control algorithms for adaptive optical systems

Yuxiang LUO1... Huizhen YANG2, Yuanfeng HE1 and Zhiguang ZHANG1,* |Show fewer author(s)

Author Affiliations

¹College of Electronic Engineering, Jiangsu Ocean University, Lianyungang 222005, China

²College of Network and Communication Engineering, Jinling Institute of Technology, Nanjing 211169, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(9)

Fig. 1. Implementation flow of hyperparameter selection using Bayesian optimization method in iterative control algorithm for adaptive optics systems

Download full size

View in Article

Fig. 2. Diagram of prediction objective function and its acquisition function after 6 iterations in SPGD algorithm

Download full size

View in Article

Fig. 3. SR curves of SPGD algorithm using traversal method and Bayesian optimization method for hyperparameter selection

Download full size

View in Article

Fig. 4. SR curves of Momentum-SPGD algorithm using traversal method and Bayesian optimization method for hyperparameter selection

Download full size

View in Article

Fig. 5. SR curves of CoolMomentum-SPGD algorithm using traversal method and Bayesian optimization method for hyperparameter selection

Download full size

View in Article

Table 1. Comparison of hyperparameter of different iterative control algorithms
View table
View in Article
Table 1. Comparison of hyperparameter of different iterative control algorithms
Algorithm Param1 Param2
SPGD $γ$
Momentum-SPGD $l_{r}$ $ρ$
CoolMomentum-SPGD $l_{r}$ $ρ_{0}$

Table 2. Comparison of results of SPGD algorithm using traversal method and Bayesian optimization method for hyperparameter selection

View table

View in Article

Table 2. Comparison of results of SPGD algorithm using traversal method and Bayesian optimization method for hyperparameter selection

Method	$γ$	Number of sample instances	Number of iterations when SR value reaches 0.8	Value of SR after 1000 iterations
Grid search	1.8	60	230	0.9213
BO_UCB	1.723	6	223	0.9244
BO_PI	1.939	7	219	0.9110
BO_EI	1.916	9	212	0.9138
BO_GP	2.0	10	206	0.9014

Table 3. Comparison of results of Momentum-SPGD algorithm using traversal method and Bayesian optimization method for hyperparameter selection

View table

View in Article

Table 3. Comparison of results of Momentum-SPGD algorithm using traversal method and Bayesian optimization method for hyperparameter selection

Method	$ρ$	$l_{r}$	Number of sample instances	Number of iterations when SR value reaches 0.8	Value of SR after 1000 iterations
Grid search	0.4	0.9	100	240	0.9336
BO_UCB	0.3409	1	7	233	0.9334
BO_PI	0.3461	0.9987	9	233	0.9334
BO_EI	0.2062	0.9935	11	270	0.9274
BO_GP	1	0.4837	11	207	0.9175

Table 4. Comparison of results of CoolMomentum-SPGD algorithm using traversal method and Bayesian optimization method for hyperparameter selection

View table

View in Article

Table 4. Comparison of results of CoolMomentum-SPGD algorithm using traversal method and Bayesian optimization method for hyperparameter selection

Method	$l_{r}$	$ρ_{0}$	Number of sample instances	Number of iterations when SR value reaches 0.8	Value of SR after 1000 iterations
Grid search	0.8	0.7	100	207	0.9363
BO_UCB	1	0.6551	9	205	0.9372
BO_PI	1	0.7130	8	228	0.9331
BO_EI	1	0.6016	8	220	0.9374
BO_GP	0.9976	0.6239	12	208	0.9380

Tools

Get Citation

Copy Citation Text

Yuxiang LUO, Huizhen YANG, Yuanfeng HE, Zhiguang ZHANG. Hyper-parameter optimization of iterative control algorithms for adaptive optical systems[J]. Journal of Applied Optics, 2024, 45(1): 126

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Research Articles

Received: Apr. 27, 2023

Accepted: --

Published Online: May. 28, 2024

The Author Email: ZHANG Zhiguang (张之光（1988—）)

DOI:10.5768/JAO202445.0102007

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Comparison of hyperparameter of different iterative control algorithms

Table 1. Comparison of hyperparameter of different iterative control algorithms

Table 2. Comparison of results of SPGD algorithm using traversal method and Bayesian optimization method for hyperparameter selection

Table 2. Comparison of results of SPGD algorithm using traversal method and Bayesian optimization method for hyperparameter selection

Table 3. Comparison of results of Momentum-SPGD algorithm using traversal method and Bayesian optimization method for hyperparameter selection

Table 3. Comparison of results of Momentum-SPGD algorithm using traversal method and Bayesian optimization method for hyperparameter selection

Table 4. Comparison of results of CoolMomentum-SPGD algorithm using traversal method and Bayesian optimization method for hyperparameter selection

Table 4. Comparison of results of CoolMomentum-SPGD algorithm using traversal method and Bayesian optimization method for hyperparameter selection

微信扫一扫：分享