人工智能教学资料 4ai Classification part2.pdfVIP

Artificial Intelligence Classification Yanghui Rao Assistant Prof., Ph.D School of Mobile Information Engineering, Sun Yat-sen University Artificial Intelligence: Classification Evaluation Metrics ? Precision – P; Recall – R ? P = a/(a+b) ? R = a/(a+c) ? F-Measure Human True False Classifier Yes a b No c d 1 1 1 1 (1 ) 2 F P R PR F P R ? ? ? ? ? ? ? Artificial Intelligence: Classification Na?ve Bayesian Classifier ? A statistical classifier ? Based on Bayes’ Theorem to perform probabilistic prediction, i.e., predict class membership probabilities ? Assumption ? The effect of an attribute on a given class is independent of other attributes ? Performance ? Comparable with decision trees and selected neural network classifiers Artificial Intelligence: Classification Na?ve Bayesian Classifier ? Let D be a training set of tuples and their associated class labels, and each tuple is represented by an n-D attribute vector X = (x1, x2, …, xn) ? Suppose there are m classes, i.e., C1, C2, …, Cm. ? Na?ve Bayesian Classifier is to derive the maximum posteriori, i.e., the maximal P(Ci|X) Artificial Intelligence: Classification Na?ve Bayesian Classifier ? This can be derived from Bayes’ theorem ? Since P(X) is constant for all classes, only needs to be maximized ? P(Ci) can be obtained from training data set si/s ( | ) ( ) ( | ) ( ) i i i P C P C P C P ? X X X ( | ) ( | ) ( )i i iP C P C P C?X X Artificial Intelligence: Classification Derivation ? Assumption: attributes are conditionally independent (i.e., no dependence relation between attributes): ? This greatly reduces the computation cost: Only counts the class distribution ? If Ak is categorical, P(xk|Ci) = sik/si, count the distribution ? If Ak is continuous-valued, P(xk|Ci) can be computed based on Gaussian distribution 1 ( | ) ( | ) n i k i k P C P x C ? ??X Artificial Intelligence: Classification Example Artificial Intelligence: Classification Comments ? Advantages ? Easy to implement ? Good results obtained in most of th