ECE 5424: Introduction to Machine Learning Topics: (Finish) Regression Model selection, Cross-validation Error decomposition Readings: Barber 17.1, 17.2 Stefan Lee Virginia Tech
Administrative Project Proposal Due: Fri 09/23, 11:55 pm NOTE: DEADLINE SHIFTED <=2pages, NIPS format HW2 Due: Wed 09/28, 11:55pm Implement linear regression, Naïve Bayes, Logistic Regression Reminder: Participation on Scholar forum is part of your grade Ask questions if you have them! (C) Dhruv Batra 2
Recap of last time (C) Dhruv Batra 3
Regression (C) Dhruv Batra 4
(C) Dhruv Batra Slide Credit: Greg Shakhnarovich 5
(C) Dhruv Batra Slide Credit: Greg Shakhnarovich 6
(C) Dhruv Batra Slide Credit: Greg Shakhnarovich 7
(C) Dhruv Batra Slide Credit: Greg Shakhnarovich 8
But, why? Why sum squared error??? Gaussians, Watson, Gaussians (C) Dhruv Batra 9
(C) Dhruv Batra Slide Credit: Greg Shakhnarovich 10
Is OLS Robust? Demo http://www.calpoly.edu/~srein/statdemo/all.html Bad things happen when the data does not come from your model! How do we fix this? (C) Dhruv Batra 11
Robust Linear Regression y ~ Lap(w x, b) On paper 5 4.5 4 L2 L1 huber 4 3 2 least squares laplace Linear data with noise and outliers 3.5 1 3 2.5 2 1.5 0 1 2 1 3 0.5 4 0 5 0.5 3 2 1 0 1 2 3 6 0 0.2 0.4 0.6 0.8 1 (C) Dhruv Batra 12
Plan for Today (Finish) Regression Bayesian Regression Different prior vs likelihood combination Polynomial Regression Error Decomposition Bias-Variance Cross-validation (C) Dhruv Batra 13
Robustify via Prior Ridge Regression y ~ N(w x, σ 2 ) w ~ N(0, t 2 I) P(w x,y) = (C) Dhruv Batra 14
Summary Likelihood Prior Name Gaussian Uniform Least Squares Gaussian Gaussian Ridge Regression Gaussian Laplace Lasso Laplace Uniform Robust Regression Student Uniform Robust Regression (C) Dhruv Batra 15
(C) Dhruv Batra Slide Credit: Greg Shakhnarovich 16
(C) Dhruv Batra Slide Credit: Greg Shakhnarovich 17
(C) Dhruv Batra Slide Credit: Greg Shakhnarovich 18
(C) Dhruv Batra Slide Credit: Greg Shakhnarovich 19
Example Demo http://www.princeton.edu/~rkatzwer/polynomialregression/ (C) Dhruv Batra 20
What you need to know Linear Regression Model Least Squares Objective Connections to Max Likelihood with Gaussian Conditional Robust regression with Laplacian Likelihood Ridge Regression with priors Polynomial and General Additive Regression (C) Dhruv Batra 21
New Topic: Model Selection and Error Decomposition (C) Dhruv Batra 22
Example for Regression Demo http://www.princeton.edu/~rkatzwer/polynomialregression/ How do we pick the hypothesis class? (C) Dhruv Batra 23
Model Selection How do we pick the right model class? Similar questions How do I pick magic hyper-parameters? How do I do feature selection? (C) Dhruv Batra 24
Errors Expected Loss/Error Training Loss/Error Validation Loss/Error Test Loss/Error Reporting Training Error (instead of Test) is CHEATING Optimizing parameters on Test Error is CHEATING (C) Dhruv Batra 25
(C) Dhruv Batra Slide Credit: Greg Shakhnarovich 26
(C) Dhruv Batra Slide Credit: Greg Shakhnarovich 27
(C) Dhruv Batra Slide Credit: Greg Shakhnarovich 28
(C) Dhruv Batra Slide Credit: Greg Shakhnarovich 29
(C) Dhruv Batra Slide Credit: Greg Shakhnarovich 30
Typical Behavior a (C) Dhruv Batra 31
Overfitting Overfitting: a learning algorithm overfits the training data if it outputs a solution w when there exists another solution w such that: (C) Dhruv Batra Slide Credit: Carlos Guestrin 32
Error Decomposition Reality (C) Dhruv Batra 33
Error Decomposition Reality (C) Dhruv Batra 34
Error Decomposition Reality Higher-Order Potentials (C) Dhruv Batra 35
Error Decomposition Approximation/Modeling Error You approximated reality with model Estimation Error You tried to learn model with finite data Optimization Error You were lazy and couldn t/didn t optimize to completion (Next time) Bayes Error Reality just sucks (C) Dhruv Batra 36