Gradient Boosting Machine is a machine learning model for classification and regression problems. In the following, we present a Condensed Gradient Boosting model that works well for multiclass multi-output regression with high precision and speed.
Gradient Boosted Neural Network-Multi-Output or GBNN-MO model is a novel training procedure for shallow and deep neural networks. The GBNN-MO is specifically developed for single and multi-output regression problems. The GBNN-MO is developed over the GBNN model.
The focus of this project is to reproduce the two well-known multioutput regression models. The SST and ERC are introduced by [pyromitros-Xioufis, Eleftherios, et al. 2016]. The main work is presented in Java here.
The main idea of this simple repository is to build an ensemble of the Multi-layer perceptrons (classifier and regressor) with the AdaBoost boosting approach.
baseline for the "Multi-target regression via input space expansion: treating targets as inputs" (Spyromitros-Xioufis, Eleftherios, et al.) experiments. Having a baseline for the reference experiment brings more insight into the advantages of using output as an input in the model.
