《智能控制智能控制智能控制 (49).pdf》由会员分享,可在线阅读,更多相关《智能控制智能控制智能控制 (49).pdf(15页珍藏版)》请在得力文库 - 分享文档赚钱的网站上搜索。
1、Methods of Machine Learning Support Vector Machine(SVM)Support Vector Machine(SVM)is a kind of generalized linear classifier that classifies data according to supervised learning.SVM s decision boundary is the maximum margin hyperplane for learning samplesMethods of Machine Learning Support Vector M
2、achine(SVM)X2X1Z2Z1Z2 Z1Methods of Machine Learning Support Vector Machine(SVM)SVM algorithm is the best algorithm between simple algorithm and neural network.Advantages:not easy to over fit,.Disadvantages:large amount of calculationIt applied to spam recognition,face recognition and other problems.
3、Methods of Machine Learning Decision Tree(DT)Decision tree algorithm is a basic classification and regression algorithm,itdivides the input space into different regions,and each region has independentparameters.Methods of Machine Learning Decision Tree(DT)Decision tree learning usually includes thre
4、e steps:Feature SelectionDecisionTree GenerationDecisionTree Pruning Methods of Machine Learning Decision Tree(DT)The following is a simple decision tree model:Methods of Machine Learning Decision Tree(DT)Advantages:simple structure,high data processing efficiency,strong interpretability and visuali
5、zation.Disadvantages:easy to over fit,difficult to tune,Unsatisfactory accuracyMethods of Machine LearningRandom ForestsRandom forest is a classifier containing multiple decision trees,Its output category is determined by the mode of the category output by individual treesMethods of Machine Learning
6、 Random ForestsRandom forest can alleviate the over fitting problem of decision tree and improve the accuracy.Random forest allows multiple algorithms to gather together equally.Introduce randomness.The random return sampling with samples is used as the input of each decision tree training sample.Me
7、thods of Machine LearningAdvantages:Good performance on classification problemFast training speedEasy to parallelizeAbility to process high-dimensional dataDisadvantages:Cant perform well in solving regression problems Cant control the operation inside the model.Depends on the amount of data.Slow ex
8、ecuting speed.Random ForestsMethods of Machine LearningNaive Bayesian Model(NBM)Naive Bayesian classification(NBC)is a method based on Bayesiantheorem and assuming that the feature conditions are independent ofeach other.CX4X3X2X1Methods of Machine Learning Naive Bayesian Model(NBM)Advantages:Simple
9、 logic,Stable algorithm,Robust performance.Disadvantages:The independence of dataset attributes is difficult to meet,Methods of Machine Learning Naive Bayesian Model(NBM)Naive Bayesian algorithm is widely used:text classification,credit evaluation,and so on spam classification phishing website detection If we have a group of samples to classify,which method mentioned above do you think is the most appropriate?Find more machine learning methods.Summary