当前位置: X-MOL 学术Pattern Recogn. Lett. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
MAGAN: A masked autoencoder generative adversarial network for processing missing IoT sequence data
Pattern Recognition Letters ( IF 3.9 ) Pub Date : 2020-07-18 , DOI: 10.1016/j.patrec.2020.07.025
Wang Weihan

Missing sequence data prevent local data from reflecting the overall distribution of a sample, hindering data analysis. The problem of missing data during actual production is a serious issue and results in a high defect rate, low dimensionality, and high noise level. In this study, a Masked Generative Adversarial Network (MAGAN) model is proposed that is less affected by the data loss rate than a baseline comparison model, and at an 80% missing data rate, the model can still better reflect the distribution of real data. MAGAN shows better results than a traditional processing method for dealing with missing data.



中文翻译:

MAGAN:用于处理丢失的IoT序列数据的蒙版自动编码器生成对抗网络

缺少序列数据会阻止本地数据反映样品的总体分布,从而妨碍数据分析。在实际生产期间丢失数据的问题是一个严重的问题,并导致高缺陷率,低尺寸和高噪声水平。在这项研究中,提出了一种蒙代式生成对抗网络(MAGAN)模型,该模型比基线比较模型受数据丢失率的影响较小,并且在丢失率达到80%的情况下,该模型仍可以更好地反映真实数据的分布。与处理丢失数据的传统处理方法相比,MAGAN显示出更好的结果。

更新日期:2020-07-28
down
wechat
bug