Lecture Slides for INTRODUCTION TO Machine Learning 2
- Slides: 24
Lecture Slides for INTRODUCTION TO Machine Learning 2 nd Edition ETHEM ALPAYDIN © The MIT Press, 2010 alpaydin@boun. edu. tr http: //www. cmpe. boun. edu. tr/~ethem/i 2 ml 2 e
CHAPTER 10: Linear Discrimination
Likelihood- vs. Discriminant-based Classification �Likelihood-based: Assume a model for p(x|Ci), use Bayes’ rule to calculate P(Ci|x) gi(x) = log P(Ci|x) �Discriminant-based: Assume a model for gi(x|Φi); no density estimation �Estimating the boundaries is enough; no need to accurately estimate the densities inside the boundaries Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 3
Linear Discriminant �Linear discriminant: �Advantages: �Simple: O(d) space/computation �Knowledge extraction: Weighted sum of attributes; positive/negative weights, magnitudes (credit scoring) �Optimal when p(x|Ci) are Gaussian with shared cov matrix; useful when classes are (almost) linearly separable Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 4
Generalized Linear Model �Quadratic discriminant: �Higher-order (product) terms: Map from x to z using nonlinear basis functions and use a linear discriminant in z-space Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 5
Two Classes Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 6
Geometry Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 7
Multiple Classes are linearly separable Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 8
Pairwise Separation Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 9
From Discriminants to Posteriors When p (x | Ci ) ~ N ( μi , ∑) Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 10
Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 11
Sigmoid (Logistic) Function Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 12
Gradient-Descent �E(w|X) is error with parameters w on sample X w*=arg minw E(w | X) �Gradient-descent: Starts from random w and updates w iteratively in the negative direction of gradient Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 13
Gradient-Descent E (wt) E (wt+1) wt η wt+1 Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 14
Logistic Discrimination �Two classes: Assume log likelihood ratio is linear Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 15
Training: Two Classes Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 16
Training: Gradient-Descent Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 17
Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 18
100 10 Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 19
K>2 Classes softmax Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 20
Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 21
Example Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 22
Generalizing the Linear Model � Quadratic: � Sum of basis functions: where φ(x) are basis functions � Hidden units in neural networks (Chapters 11 and 12) � Kernels in SVM (Chapter 13) Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 23
Discrimination by Regression �Classes are NOT mutually exclusive and exhaustive Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2 e © The MIT Press (V 1. 0) 24
- Cmu machine learning
- Introduction to machine learning slides
- Introduction to machine learning slides
- Machine learning lecture notes
- A small child slides down the four frictionless slides
- John pushes hector on a plastic toboggan
- Principles of economics powerpoint lecture slides
- Business communication lecture slides
- 01:640:244 lecture notes - lecture 15: plat, idah, farad
- Concept learning task in machine learning
- Analytical learning in machine learning
- Pac learning model in machine learning
- Pac learning model in machine learning
- Inductive and analytical learning in machine learning
- Inductive learning approach
- Instance based learning in machine learning
- Inductive learning machine learning
- First order rule learning in machine learning
- Lazy learners vs eager learner
- Deep learning vs machine learning
- Introduction to machine learning ethem alpaydin
- Andrew ng introduction to machine learning
- Andrew ng introduction to machine learning
- Mike mozer
- Introduction to machine learning ethem alpaydin