机器学习 吴恩达 Lecture6.pptx

LogisticRegressionClassificationMachine LearningClassificationEmail: Spam / Not Spam?Online Transactions: Fraudulent (Yes / No)?Tumor: Malignant / Benign ?0: “Negative Class” (e.g., benign tumor) 1: “Positive Class” (e.g., malignant tumor)(Yes) 1Malignant ?(No) 0Tumor SizeTumor SizeThreshold classifier output at 0.5:If , predict “y = 1”If , predict “y = 0”Classification: y = 0 or 1Logistic Regression:can be 1 or 0LogisticRegressionHypothesisRepresentationMachine LearningLogistic Regression ModelWant10.5Sigmoid functionLogistic function0Interpretation of Hypothesis Output= estimated probability that y = 1 on input x Example: If Tell patient that 70% chance of tumor being malignant “probability that y = 1, given x, parameterized by ”LogisticRegressionDecision boundaryMachine LearningLogistic regression1z Suppose predict “ “ if predict “ “ ifDecision Boundaryx2321x1123Predict “ “ if Non-linear decision boundariesx21x11-1Predict “ “ if -1x2x1LogisticRegressionCost functionMachine LearningTraining set:m examplesHow to choose parameters ?Cost functionLinear regression:“non-convex”“convex”Logistic regression cost functionIf y = 101Logistic regression cost functionIf y = 001LogisticRegressionSimplified cost function and gradient descentMachine LearningLogistic regression cost functionLogistic regression cost functionTo fit parameters : To make a prediction given new :Output Gradient DescentWant :Repeat(simultaneously update all )Gradient DescentWant :Repeat(simultaneously update all )Algorithm looks identical to linear regression!LogisticRegressionAdvanced optimizationMachine LearningOptimization algorithmCost function . Want .Given , we have code that can compute (for )Gradient descent:RepeatOptimization algorithmGiven , we have code that can compu