A feature reconstruction-based multi-task regression model for cyanobacterial distribution forecasting along the water column
A feature reconstruction-based multi-task regression model for cyanobacterial distribution forecasting along the water column
复制标题
基于特征重建的多任务回归模型用于蓝藻沿水体分布预测
DOI：
10.1016/j.jclepro.2021.126025
复制
发表时间：
2021-04
影响因子：
11.1
通讯作者：
Karina Yew-Hoong Gin
中科院分区：
文献类型：
--
作者：
Peng Jiang;Yibing Huang;Xiao LIU;Jingjie Zhang;Karina Yew-Hoong Gin
蓝藻污染对水生态系统和人类健康的危害已经威胁到清洁的生态系统和城市的可持续发展，因此需要开发一种有效的蓝藻水华预测工具。蓝藻细胞密度沿水体沿着的变化呈现出不同的分布规律，并受多种环境因子的影响。大多数数据驱动模型将特定水深的蓝藻预测视为单一任务，无法在水深之间共享知识，导致预测精度不高。这就是为什么越来越多的非线性黑箱模型已建成蓝藻预测，但在模型的可解释性的代价。本研究的目的是调查是否可以提高预测精度和模型的可解释性（i）使用易于访问的预测和（ii）开发一个功能重建为基础的多任务回归模型与知识共享之间的水深。来自热带湖泊的真实数据被用来评估模型的有效性。对于所研究的湖泊，最高的平均蓝藻细胞密度出现在1.0米，之后，它减少了30%以上，在5.5米。相邻水深蓝藻细胞密度的时间序列相关系数均大于0.95（P < 0.001）。预测结果表明，与单任务非线性模型相比，提前一天，提前两天，提前三天蓝藻水华预测的误差减少20.59%，16.25%和22.70%，测量的均方误差。在该模型下，水华和非水华信号的准确率分别达到94.81%和98.28%。基于该模型，预测因子的相对重要性、回归系数的稀疏性以及回归系数的协方差关系可以充分解释模型，并阐明知识共享和预测精度提高的机理。
Cyanobacterial water pollution has been threatening the cleaner ecosystem and urban sustainability due to the harmfulness to aquatic ecosystems and human health, which triggers the development of an effective forecasting tool for cyanobacterial blooms. Along the water column, the variations in cyanobacteria cell densities show various distribution patterns and are influenced by multiple environmental factors. Most data-driven models treat cyanobacteria forecasting at a specific water depth as a single task, which fails to share knowledge amongst water depths, resulting in unfavourable forecasting accuracy. This is why an increasing number of nonlinear black-box models have been built for cyanobacteria forecasting but at the expense of model interpretability. This study aims to investigate whether forecasting accuracy and model interpretability can be enhanced by (i) using easily accessible predictors and (ii) developing a feature reconstruction-based multi-task regression model with knowledge sharing amongst water depths. Real-world data from a tropical lake are used to evaluate the effectiveness of the model. For the studied lake, the highest average cyanobacteria cell density occurs at 1.0 m, after which it decreases by over 30% at 5.5 m. The correlation coefficients of time-serial cyanobacteria cell densities between adjacent water depths are greater than 0.95 (P < 0.001). The forecasting results indicate that, compared to single-task nonlinear models, 20.59%, 16.25%, and 22.70% error reductions, measured by the mean square error, are achieved for one-day-ahead, two-day-ahead, and three-day-ahead cyanobacterial bloom forecasts. The accurate bloom and non-bloom signals under the proposed model are up to 94.81% and 98.28%. Based on the proposed model, the relative importance of predictors, the sparsity of regression coefficients, and the covariance relationship of regression coefficients can interpret the model adequately and elucidate the mechanism of knowledge sharing and forecasting accuracy improvement.
登录
查看更多内容
影响因子：
11.1
作者：
Nuamah Linda A.;Li Yiping;Pu Yashuai;Nwankwegu Amechi S.;Haikuo Zhang;Norgbey Eyram;Banahene Patrick;Bofah-Buoh Robert
通讯作者：
Bofah-Buoh Robert
DOI：
10.1016/j.envsoft.2014.08.014
发表时间：
2014-12
期刊：
Environ. Model. Softw.
影响因子：
--
作者：
Yeesock Kim;Hyun-suk Shin;J. Plummer
通讯作者：
Yeesock Kim;Hyun-suk Shin;J. Plummer
DOI：
10.1007/978-3-540-87527-7_21
发表时间：
2008-09
期刊：
IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics)
影响因子：
--
作者：
Zhi-hui Zhan;Jun Zhang
通讯作者：
Zhi-hui Zhan;Jun Zhang
影响因子：
3.1
作者：
F. Mattei;Simone Franceschini;M. Scardi
通讯作者：
F. Mattei;Simone Franceschini;M. Scardi
DOI：
10.1039/c1em10127g
发表时间：
2011-06
期刊：
Journal of environmental monitoring : JEM
影响因子：
--
作者：
R. Torres;E. Pereira;V. Vasconcelos;L. O. Teles
通讯作者：
R. Torres;E. Pereira;V. Vasconcelos;L. O. Teles