首页|Hohai University Reports Findings in Machine Learning [Uncert ainty-based saltwater intrusion prediction using integrated Bayesian machine lea rning modeling (IBMLM) in a deep aquifer]
Hohai University Reports Findings in Machine Learning [Uncert ainty-based saltwater intrusion prediction using integrated Bayesian machine lea rning modeling (IBMLM) in a deep aquifer]
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-New research on Machine Learning is th e subject of a report. According to news reporting from Nanjing, People's Republ ic of China, by NewsRx journalists, research stated, "Data-driven machine learni ng approaches are promising to substitute physically based groundwater numerical models and capture input-output relationships for reducing computational burden . But the performance and reliability are strongly influenced by different sourc es of uncertainty." The news correspondents obtained a quote from the research from Hohai University , "Conventional researches generally rely on a stand-alone machine learning surr ogate approach and fail to account for errors in model outputs resulting from st ructural deficiencies. To overcome this issue, this study proposes a flexible in tegrated Bayesian machine learning modeling (IBMLM) method to explicitly quantif y uncertainties originating from structures and parameters of machine learning s urrogate models. An Expectation-Maximization (EM) algorithm is combined with Bay esian model averaging (BMA) to find out maximum likelihood and construct posteri or predictive distribution. Three machine learning approaches representing diffe rent model complexity are incorporated in the framework, including artificial ne ural network (ANN), support vector machine (SVM) and random forest (RF). The pro posed IBMLM method is demonstrated in a field-scale real-world ‘1500-foot' sand aquifer, Baton Rouge, USA, where overexploitation caused serious saltwater intru sion (SWI) issues. This study adds to the understanding of how chloride concentr ation transport responds to multi-dimensional extraction-injection remediation s trategies in a sophisticated saltwater intrusion model. Results show that most I BMLM exhibit r values above 0.98 and NSE values above 0.93, both slightly higher than individual machine learning, confirming that the IBMLM is well established to provide better model predictions than individual machine learning models, wh ile maintaining the advantage of high computing efficiency. The IBMLM is found u seful to predict saltwater intrusion without running the physically based numeri cal simulation model. We conclude that an explicit consideration of machine lear ning model structure uncertainty along with parameters improves accuracy and rel iability of predictions, and also corrects uncertainty bounds."
NanjingPeople's Republic of ChinaAsi aCyborgsEmerging TechnologiesMachine Learning
NETL
NSTL
