Robust Anomaly Recognition in Hydraulic Structural Safety Monitoring: A Methodology Based on Deconfounding Boosted Regression Trees

Algorithm 2

Training of a copula debiased boosted regression trees model.

Input: Train data collection , correlation coefficient ρ, loss function L, max iteration number M, and learning rate λ

Output: Regression function 1 //Initialize regression function2whiledo3   //Get residual’s gradient of 4  5  //Train CART tree with bootstrap method and sample residual’s standard error 6  7  8  9  //Train a new CART tree to update regression function of 10end11Return