首页|Investigators from Shanghai University Report New Data on Intelligent Systems (A Multi-task Learning-based Generative Adversarial Network for Red Tide Multivari ate Time Series Imputation)
Investigators from Shanghai University Report New Data on Intelligent Systems (A Multi-task Learning-based Generative Adversarial Network for Red Tide Multivari ate Time Series Imputation)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News Daily News-Research findings on Machine Learning - Intelligent Systems are discussed in a new report. According to news reporting from Shanghai, People's Republic of China, by NewsRx journalists, research stat ed, "Red tide data are typical multivariate time series (MTS) and complete data help analyze red tide more conveniently. However, missing values due to artifici al or accidental events hinder further analysis of red tide phenomenon." The news correspondents obtained a quote from the research from Shanghai Univers ity, "Generative adversarial network (GAN) is effective in capturing distributio n of MTS while the imputation performance is far from satisfactory, especially i n conditions of high missing rate. One of the remaining open challenges is that common GAN-based imputation methods usually lack the ability to excavate implici t correlations between different attributions and downstream tasks, from which a dvanced latent information about missing values can be mined to improve imputati on performance. To deal with the problem, a novel multi-task learning-based gene rative adversarial imputation network (MTGAIN) is proposed by introducing the pr ediction task into GAN to unearth more detailed information about missing values to better model distribution of red tide MTS. Furthermore, the homoscedastic un certainty of multiple tasks is exploited to balance the weights of losses betwee n generation and prediction tasks."
ShanghaiPeople's Republic of ChinaAs iaIntelligent SystemsMachine LearningShanghai University
NETL
NSTL
