Boosting Based Conditional Quantile Estimation for Regression and Binary Classification

We introduce Quantile Boost (QBoost) algorithms which predict conditional quantiles of the interested response for regression and binary classification. Quantile Boost Regression (QBR) performs gradient descent in functional space to minimize the objective function used by quantile regression (QReg). In the classification scenario, the class label is defined via a hidden variable, and the quantiles of the class label are estimated by fitting the corresponding quantiles of the hidden variable. An equivalent form of the definition of quantile is introduced, whose smoothed version is employed as the objective function, which is maximized by gradient ascent in functional space to get the Quantile Boost Classification (QBC) algorithm. Extensive experiments show that QBoost performs better than the original QReg and other alternatives for regression and classification. Furthermore, QBoost is more robust to noisy predictors.