Federated conditional generative adversarial nets imputation method for air quality missing data

作者：

Highlights：

• New method for air quality missing data imputation.

• A new task of imputation for the missing data of air quality. This paper considers that the air quality monitoring data belongs to several different data holders without sharing.

• This paper, for the first time, proposes a GAN-based imputation method with federated learning framework for missing data to fix the missing data.

• We implemented the methods using Pytorch and open our project codes on Github.

摘要

•New method for air quality missing data imputation.•A new task of imputation for the missing data of air quality. This paper considers that the air quality monitoring data belongs to several different data holders without sharing.•This paper, for the first time, proposes a GAN-based imputation method with federated learning framework for missing data to fix the missing data.•We implemented the methods using Pytorch and open our project codes on Github.

论文关键词：Air pollutants,Conditional GAN imputation,Federated learning,Privacy-preserving machine learning

论文评审过程：Received 10 March 2021, Revised 24 May 2021, Accepted 24 June 2021, Available online 26 June 2021, Version of Record 26 July 2021.

论文官网地址：https://doi.org/10.1016/j.knosys.2021.107261