Outline Classification Linear classifiers Perceptron Multi-class classification Generative approach Naïve Bayes classifier 2
Classification: Oranges and Lemons 3
4
Classification problem 5
Linear classifiers 6
7
Decision boundary 8
Linear Decision boundary (Perceptron) 9
Linear Decision boundary (Decision Tree) 10 t1t3 t2 Income
Linear Decision boundary (K Nearest Neighbor) 11 O O O x x x Feature 1 Feature 2
Non-Linear Decision boundary 12 Decision Boundary Decision Region 1 Decision Region 2
Decision boundary Linear classifier 13
Non-linear decision boundary Choose non-linear features Classifier still linear in parameters 14
Linear boundary: geometry 15
SSE cost function for classification SSE cost function is not suitable for classification Sum of Squared Errors loss penalizes “too correct” predictions SSE also lack robustness to noise 16
SSE cost function for classification 17
Perceptron algorithm 18
Perceptron criterion 19
Batch gradient for descent Perceptron “Gradient Descent” to solve the optimization problem Batch Perceptron converges in finite number of steps for linearly separable data 20
Stochastic gradient descent for Perceptron 21
Convergence of Perceptron 22
Convergence of Perceptron 23
Multi-class classification 24
Multi-class classification One-vs-all (one-vs-rest) 25
Multi-class classification One-vs-one 26
Multi-class classification: ambiguity regions in which the classification is undefined Converting the multi-class problem to a set of two- class problems can lead to regions in which the classification is undefined 27
Probabilistic approach Bayes’ theorem 28
Bayes’ theorem 29
Bayes decision theory 30
Probabilistic classifiers Probabilistic classification approaches can be divided in two main categories Generative Discriminative 31
Discriminative vs. generative approach 32
Generative approach 33
Discriminative approach 34
Naïve Bayes classifier 35
Naïve Bayes classifier 36
Naïve Bayes: discrete example 37
38