XGBoost: its Genealogy, its Architectural Features, and its Innovation

Sequential and Parallel Architectures All in One

INTRODUCTION

XGBoost (eXtreme Gradient Boost) is a powerful learning algorithm which had outperformed many conventional Machine Learning algos in many competitions in the past.

In a nutshell, XGBoost is equipped with both sequential and parallel architectures all in one: while it is a sequential learning algorithm (additive strategy), it incorporates parallel computation into its architecture in order to enhance the system efficiency.

bias-variance-tradeoff ensemble-learning features genealogy innovation machine learning newton-raphson-method xgboost