What is gradient boosting and how does it differ from traditional boosting methods?

Gradient Boosting is an ensemble learning technique used for regression and classification that builds models sequentially. In each iteration, it fits a new model to the residual errors made by the previous models, thereby optimizing a loss function. Unlike traditional boosting methods like AdaBoost, which adjusts the weights of misclassified observations, Gradient Boosting focuses on the gradients of the loss function to minimize them over iterations. This allows for more flexibility and tuning, making it generally more powerful and applicable to a wider range of problems, but it can also be more prone to overfitting without proper regularization.