陈天奇论文演讲课件.pptVIP

陈天奇论文演讲;Outline;Elements in Supervised Learning;Elements continued: Objective Function;Putting known knowledge into context;Objective and Bias Variance Trade-off;Outline;Regression Tree (CART);;Tree Ensemble methods;Put into context: Model and Parameters;Learning a tree on single variable;Learning a step function;Learning step function (visually);Coming back: Objective for Tree Ensemble;Objective vs Heuristic;Regression Tree is not just for regression!;Outline;Take Home Message for this section;So How do we Learn?;Additive Training;Taylor Expansion Approximation of Loss;Our New Goal;Refine the definition of tree;Define Complexity of a Tree (cont’);Revisit the Objectives;The Structure Score;The Structure Score Calculation;Searching Algorithm for Single Tree;Greedy Learning of the Tree;Efficient Finding of the Best Split;An Algorithm for Split Finding;What about Categorical Variables?;Pruning and Regularization;Recap: Boosted Tree Algorithm;Outline;Questions to check if you really get it;Questions to check if you really get it;Questions to check if you really get it;Summary;Reference