Analysis and Mitigation of Unwanted Biases in ML-based QoT Classification Tasks.

Carlos Natalino,Behnam Shariati,Pooyan Safari,Johannes Karl Fischer, Paolo Monti

Optical Fiber Communications Conference and Exhibition（2024）

引用 0|浏览0

暂无评分

摘要

We address the problem of mitigating biases in models used for the quality of transmission prediction. The proposed method reduces the relative accuracy difference between samples with different feature values by up to 45%.

查看译文

关键词

Unwanted Bias,Feature Values,Artificial Neural Network,Sampling Weights,Data Bias,Binary Cross-entropy Loss,Balanced Dataset,Optical Networks,Standard Deviation Decrease,Modulation Formats,Artificial Intelligence Machine Learning

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