Study On Optical Communications, Volume. 48, Issue 6, 45(2022)
Transfer Learning Assisted Transmission Quality Evaluation based on Machine Learning
Figures & Tables(9)
Fig. 2. The structure diagram of migration and fine-tuning classifier
Table 1. A, B and C link parameter settings
View tableView in ArticleTable 1. A, B and C link parameter settings
参数 A参数范围 B参数范围 C参数范围 链路长度/km 80～7 500 90～7 500 50～7 000 跨度长度/km 80，100，120，150 90，110，130，150 50，80，110，140 调制格式 PM-BPSK，PM-QPSK，16QAM，64QAM PM-BPSK，PM-QPSK，16QAM，64QAM PM-BPSK，PM-QPSK，16QAM，64QAM 数据速率Gbit/s 40，50，100 40，50，100 50，100 信道发射功率/dBm -10～5 -10～6 -15～0
Table 2. Dataset content
View tableView in ArticleTable 2. Dataset content
总样本数量 类别0(优良) 类别1(合格) 类别2(不合格) 数据集A 32 991 8 722 4 391 19 878 数据集B 20 375 7 318 2 512 10 545 数据集C 20 215 6 724 2 603 10 888
Table 3. Multi-classification performances of system B
View tableView in ArticleTable 3. Multi-classification performances of system B
Kappa 海明距离 准确率 F1_macro SVM 0.885 5 0.067 6 0.932 3 0.879 0 随机森林 0.824 4 0.103 6 0.896 4 0.812 6 ANN 0.880 5 0.070 5 0.929 5 0.872 7 DNN 0.881 1 0.070 1 0.929 8 0.873 5
Table 4. Multi-classification performances of system C
View tableView in ArticleTable 4. Multi-classification performances of system C
Kappa 海明距离 准确率 F1_macro SVM 0.810 3 0.111 5 0.888 5 0.831 4 随机森林 0.632 7 0.209 1 0.790 9 0.675 9 ANN 0.842 2 0.092 0 0.908 0 0.851 6 DNN 0.840 2 0.093 6 0.907 4 0.850 2
Table 5. Multi-classification performances of system B
View tableView in ArticleTable 5. Multi-classification performances of system B
Kappa 海明距离 准确率 F1_macro SVM 0.956 6 0.025 6 0.974 3 0.954 7 随机森林 0.886 9 0.066 3 0.933 7 0.880 6 ANN 0.963 0 0.021 8 0.978 2 0.962 2 DNN 0.970 1 0.017 6 0.982 4 0.968 4
Table 6. Multi-classification performances of system C
View tableView in ArticleTable 6. Multi-classification performances of system C
Kappa 海明距离 准确率 F1_macro SVM 0.953 0 0.027 5 0.972 5 0.954 2 随机森林 0.916 2 0.048 4 0.951 6 0.917 4 ANN 0.961 9 0.022 2 0.977 8 0.961 7 DNN 0.968 2 0.018 5 0.981 5 0.968 3
Tools
Get Citation
Copy Citation Text
Jia-xin WANG. Transfer Learning Assisted Transmission Quality Evaluation based on Machine Learning[J]. Study On Optical Communications, 2022, 48(6): 45
Paper Information
Category: Research Articles
Received: Apr. 7, 2022
Accepted: --
Published Online: Feb. 14, 2023
The Author Email: Jia-xin WANG (1002054577@qq.com)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20