Monitoring Data Augmentation of Spectral Information using VAE and GAN for Soft-Failure Identification.

Lars E. Kruse,Sebastian Kühl,Annika Dochhan,Stephan Pachnicke

Optical Fiber Communications Conference and Exhibition（2024）

引用 0|浏览4

暂无评分

摘要

We propose data augmentation of monitoring information using VAE and GAN to reduce the amount of required soft-failure training data. Results show that only 5 samples per failure type are needed for F1-scores above 0.9.

查看译文

关键词

Data Augmentation,Generative Adversarial Networks,Variational Autoencoder,Training Data,Type Of Failure,Neural Network,Learning Algorithms,Input Features,Input Layer,Central Focus,Classification Tree,Optical Spectra,Latent Space,Bit Error Rate,Amount Of Training Data,Optical Amplifier,Optical Spectrum Analyzer,Quality Of Transmission,Standard Single-mode Fiber,Variable Optical Attenuator

AI 理解论文

溯源树

样例

生成溯源树，研究论文发展脉络

Chat Paper

正在生成论文摘要