Introduction to machine learning(机器学习简介).ppt

* * * * * * * * * * * Multinormal~Dirichlet(1) Multinormal * Lab of Semantic Computing and Data Mining of USTC * Gaussian Distribution * Lab of Semantic Computing and Data Mining of USTC * The Category of Machine Learning Supervised learning: classification, regression Unsupervised learning: clustering, density estimation, visualization Semi-supervised learning reinforcement learning: * Lab of Semantic Computing and Data Mining of USTC * Decision Theory Minimizing the misclassification rate Minimizing the expected loss Loss function, cost function Inference and decision Generative Model : p(x,Ck) discriminative models : p(Ck|x) discriminant function: δ(C(x), Ck) Information Theory Kullback-Leibler divergence (KL divergence) Jensen’s inequality if f(x) is a strictly convex function and p(x) is a probability function * Lab of Semantic Computing and Data Mining of USTC * Nonparametric methods KNN, kernel, histogram * Lab of Semantic Computing and Data Mining of USTC * * * * * * * * * * * * * * * * * * * * * * * * * * * Lab of Semantic Computing and Data Mining of USTC Introduction to machine learning Biao Xiang * Lab of Semantic Computing and Data Mining of USTC * Contents Several Key Concepts Bayesian Theorem Properties of Matrices Model English Dictionary: something that is copied or used as the basis for a related idea, process, or system; Wikipedia: A?statistical model?is a formalization of relationships between variables in the form of mathematical equations. In my words: The assumption for the relationships between variables in your opinion. A Example: give a model for the human height heighti?= b0?+ b1agei?+ b2sexi?+ εi How to construct the model for a problem: An assumption Estimating the coefficients in the model * Lab of Semantic Computing and Data Mining of USTC * How? * Lab of Semantic Computing and Data Mining of USTC * Social Influence Model: Given a network G(V, E, T) Axiom 1: Axiom 2: Axiom 3: Model Coefficient 统计模型的理论基