Modeling Method of Miniaturized Nonlinear All-Optical Diffraction Deep Neural Network Based on 10.6 μm Wavelength

Yichen Sun; Mingli Dong; Mingxin Yu; Jiabin Xia; Xu Zhang; Yuchen Bai; Lidan Lu; Lianqing Zhu

doi:10.3788/LOP202158.0820001

Laser & Optoelectronics Progress, Volume. 58, Issue 8, 0820001(2021)

Modeling Method of Miniaturized Nonlinear All-Optical Diffraction Deep Neural Network Based on 10.6 μm Wavelength

Yichen Sun, Mingli Dong^*, Mingxin Yu^**, Jiabin Xia, Xu Zhang, Yuchen Bai, Lidan Lu, and Lianqing Zhu

Key Laboratory of the Ministry of Education for Optoelectronic Measurement Technology and Instrument, Beijing Information Science and Technology University, Beijing 100192, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(22)

Fig. 1. Dataset examples. (a) MNIST dataset; (b) Fashion-MNIST dataset

Download full size

$Structural diagrams of nonlinear all-optical diffraction deep neural network. (a) Physical model of system; (b) optical path model; (c) neural network model$

Fig. 2. Structural diagrams of nonlinear all-optical diffraction deep neural network. (a) Physical model of system; (b) optical path model; (c) neural network model

Download full size

Fig. 3. Mathematical models of different activation functions. (a) Leaky-ReLU and PReLU; (b) RReLU

Download full size

Fig. 4. Image label design

Download full size

Fig. 5. Output image of each layer of grating after training

Download full size

Fig. 6. Classification accuracy corresponding to each number of grating layers in MNIST dataset

Download full size

Fig. 7. Classification accuracy of Fashion-MNIST dataset corresponding to each number of grating layers in Fashion-MNIST dataset

Download full size

$Classification results of MNIST dataset by standard all-optical diffraction deep neural network. (a) Classification accuracy; (b) confusion matrix$

Fig. 8. Classification results of MNIST dataset by standard all-optical diffraction deep neural network. (a) Classification accuracy; (b) confusion matrix

Download full size

$Classification results of Fashion-MNIST dataset by standard all-optical diffraction deep neural network. (a) Classification accuracy; (b) confusion matrix$

Fig. 9. Classification results of Fashion-MNIST dataset by standard all-optical diffraction deep neural network. (a) Classification accuracy; (b) confusion matrix

Download full size

$Classification accuracies and confusion matrixes of MNIST dataset by all-optical diffraction deep neural networks with different activation functions. (a)(b) Leaky-ReLU; (c)(d) PReLU; (e)(f) RReLU$

Fig. 10. Classification accuracies and confusion matrixes of MNIST dataset by all-optical diffraction deep neural networks with different activation functions. (a)(b) Leaky-ReLU; (c)(d) PReLU; (e)(f) RReLU

Download full size

$Recognition accuracy of each number in MNIST dataset by each all-optical diffraction deep neural network model$

Fig. 11. Recognition accuracy of each number in MNIST dataset by each all-optical diffraction deep neural network model

Download full size

$Classification accuracies and confusion matrixes of Fashion-MNIST dataset by all-optical diffraction deep neural networks with different activation functions. (a)(b) Leaky-ReLU; (c)(d) PReLU; (e)(f) RReLU$

Fig. 12. Classification accuracies and confusion matrixes of Fashion-MNIST dataset by all-optical diffraction deep neural networks with different activation functions. (a)(b) Leaky-ReLU; (c)(d) PReLU; (e)(f) RReLU

Download full size

$Recognition accuracy of each number in Fashion-MNIST dataset by each all-optical diffraction deep neural network model$

Fig. 13. Recognition accuracy of each number in Fashion-MNIST dataset by each all-optical diffraction deep neural network model

Download full size

Table 1. Label numbers and categories in Fashion-MNIST dataset
View table
Table 1. Label numbers and categories in Fashion-MNIST dataset
Label number Category
0 T-shirt
1 Trousers
2 Pullover
3 Dress
4 Coat
5 Sandal
6 Shirt
7 Sneaker
8 Bag
9 Ankle boot

Table 2. Physical parameters of neural network grating in MNIST dataset
View table
Table 2. Physical parameters of neural network grating in MNIST dataset
Grating parameter Value
Wavelength 10.6 μm
Cell size 5 μm
Grating spacing 70λ

Table 3. Neural network training parameters in MNIST dataset
View table
Table 3. Neural network training parameters in MNIST dataset
Training parameter Value
Number of grating layers 6
Number of neurons per layer 60×60
Batch size 100
Epoch 50
Learning rate 10^-2

Table 4. Classification accuracy of MNIST dataset corresponding to each pixel size and diffraction grating spacing

View table

Table 4. Classification accuracy of MNIST dataset corresponding to each pixel size and diffraction grating spacing

Spacing	Pixel size of 30×30	Pixel size of 40×40	Pixel size of 50×50	Pixel size of 60×60	Pixel size of 70×70
30λ	0.8427	0.8642	0.8736	0.8694	0.8664
40λ	0.8218	0.8623	0.8707	0.8744	0.8667
50λ	0.7545	0.8614	0.8594	0.8759	0.8712
60λ	0.6499	0.8327	0.8710	0.8714	0.8741
70λ	0.6190	0.8304	0.8683	0.8765	0.8696

Table 5. Physical parameters of neural network grating in Fashion-MNIST dataset
View table
Table 5. Physical parameters of neural network grating in Fashion-MNIST dataset
Grating parameter Value
Wavelength 10.6 μm
Cell size 5 μm
Grating spacing 30λ

Table 6. Neural network training parameters in Fashion-MNIST dataset
View table
Table 6. Neural network training parameters in Fashion-MNIST dataset
Training parameter Value
Number of grating layers 6
Number of neurons per layer 70×70
Batch size 100
Epoch 50
Learning rate 10^-2

Table 7. Classification accuracy of Fashion-MNIST dataset corresponding to each pixel size and diffraction grating spacing

View table

Table 7. Classification accuracy of Fashion-MNIST dataset corresponding to each pixel size and diffraction grating spacing

Spacing	Pixel size of 30×30	Pixel size of 40×40	Pixel size of 50×50	Pixel size of 60×60	Pixel size of 70×70
30λ	0.7012	0.7797	0.7943	0.7969	0.7994
40λ	0.6569	0.7539	0.7882	0.7903	0.7947
50λ	0.6137	0.7419	0.7664	0.7849	0.7937
60λ	0.6098	0.7411	0.7574	0.7831	0.7935
70λ	0.6069	0.7246	0.7539	0.7735	0.7809

Table 8. Classification accuracies of MNIST dataset by nonlinear all-optical diffraction deep neural networks with different activation functions
View table
Table 8. Classification accuracies of MNIST dataset by nonlinear all-optical diffraction deep neural networks with different activation functions
Activation function Accuracy
Leaky-ReLU 0.9609
PReLU 0.9628
RReLU 0.9630

Table 9. Classification accuracies of Fashion-MNIST dataset by nonlinear all-optical diffraction deep neural networks with different activation functions
View table
Table 9. Classification accuracies of Fashion-MNIST dataset by nonlinear all-optical diffraction deep neural networks with different activation functions
Activation function Accuracy
Leaky-ReLU 0.8717
PReLU 0.8736
RReLU 0.8743

Tools

Get Citation

Copy Citation Text

Yichen Sun, Mingli Dong, Mingxin Yu, Jiabin Xia, Xu Zhang, Yuchen Bai, Lidan Lu, Lianqing Zhu. Modeling Method of Miniaturized Nonlinear All-Optical Diffraction Deep Neural Network Based on 10.6 μm Wavelength[J]. Laser & Optoelectronics Progress, 2021, 58(8): 0820001

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Optics in Computing

Received: Sep. 29, 2020

Accepted: Nov. 5, 2020

Published Online: Apr. 16, 2021

The Author Email: Mingli Dong (dongml@bistu.edu.cn), Mingxin Yu (yumingxin@bistu.edu.cn)

DOI:10.3788/LOP202158.0820001

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Label numbers and categories in Fashion-MNIST dataset

Table 1. Label numbers and categories in Fashion-MNIST dataset

Table 2. Physical parameters of neural network grating in MNIST dataset

Table 2. Physical parameters of neural network grating in MNIST dataset

Table 3. Neural network training parameters in MNIST dataset

Table 3. Neural network training parameters in MNIST dataset

Table 4. Classification accuracy of MNIST dataset corresponding to each pixel size and diffraction grating spacing

Table 4. Classification accuracy of MNIST dataset corresponding to each pixel size and diffraction grating spacing

Table 5. Physical parameters of neural network grating in Fashion-MNIST dataset

Table 5. Physical parameters of neural network grating in Fashion-MNIST dataset

Table 6. Neural network training parameters in Fashion-MNIST dataset

Table 6. Neural network training parameters in Fashion-MNIST dataset

Table 7. Classification accuracy of Fashion-MNIST dataset corresponding to each pixel size and diffraction grating spacing

Table 7. Classification accuracy of Fashion-MNIST dataset corresponding to each pixel size and diffraction grating spacing

Table 8. Classification accuracies of MNIST dataset by nonlinear all-optical diffraction deep neural networks with different activation functions

Table 8. Classification accuracies of MNIST dataset by nonlinear all-optical diffraction deep neural networks with different activation functions

Table 9. Classification accuracies of Fashion-MNIST dataset by nonlinear all-optical diffraction deep neural networks with different activation functions

Table 9. Classification accuracies of Fashion-MNIST dataset by nonlinear all-optical diffraction deep neural networks with different activation functions