CS 691A Machine Learning
Risk Minimization and Model Selection
Gianfranco Doretto
West Virginia University
http:/www.csee.wvu.edu/~gidoretto/courses/2016-fall-ml/index.html
What have we seen so far?
Several algorithms that seem to work fine on training da
CS 691A Machine Learning
Perceptron and Kernels
Gianfranco Doretto
West Virginia University
http:/www.csee.wvu.edu/~gidoretto/courses/2016-fall-ml/index.html
Outline
Perceptron
Hebbian learning
Algorithm
Convergence analysis
Features and preprocessin
CS 691A Machine Learning
Linear Regression
Gianfranco Doretto
West Virginia University
http:/www.csee.wvu.edu/~gidoretto/courses/2016-fall-ml/index.html
Outline
Optimal Regression
Linear Regression
Learning
Ordinary Least Square (MLE)
Regularized (MA
CS 691A Machine Learning
Instance Based Learning
Gianfranco Doretto
West Virginia University
http:/www.csee.wvu.edu/~gidoretto/courses/2016-fall-ml/index.html
Outline
Density Estimation
Binning
Bias-variance tradeoff
Kernel density estimation
Smoothing k
CS 691A Machine Learning
Logistic Regression
Gianfranco Doretto
West Virginia University
http:/www.csee.wvu.edu/~gidoretto/courses/2016-fall-ml/index.html
Outline
Logistic Regression
MLE Training
Gradient ascent
MAP Training
Gradient ascent
Generati
CS 691A Machine Learning
Nave Bayes
Gianfranco Doretto
West Virginia University
http:/www.csee.wvu.edu/~gidoretto/courses/2016-fall-ml/index.html
Outline
An application of Bayes rule
The Nave Bayes classifier
0-1 Loss Bayesian estimator
Nave Bayes assum
CS 691A Machine Learning
Support Vector Classification
Gianfranco Doretto
West Virginia University
http:/www.csee.wvu.edu/~gidoretto/courses/2016-fall-ml/index.html
Outline
Support Vector Classification
Large Margin Separation, optimization problem
Prop