Gradient Boosting is an ensemble learning technique used in machine learning to improve the accuracy of predictions by combining the outputs of multiple weak learners, typically decision trees, into a strong predictor.