Semester:  Fall 2016, also offered on Fall 2015 by Dan Goldwasser 
Time and place:  Tuesday and Thursday, 3.00pm4.15pm, SengLiang Wang Hall 2599 
Instructor:  Jean Honorio, Lawson Building 2142J (Please send an email for appointments) 
TAs: 
Chang Li, email: li1873 at purdue.edu, Office hours: Monday, 11am1pm, HAAS G50 Rohit Rangan, email: rrangan at purdue.edu, Office hours: Friday, 3pm5pm, HAAS G50 
Date  Topic (Tentative)  Notes 
Tue, Aug 23  Lecture 1: perceptron (introduction)  Homework 0: due on Aug 25 at beginning of class  NO EXTENSION DAYS ALLOWED 
Thu, Aug 25  Lecture 2: perceptron (convergence), maxmargin classifiers, support vector machines (introduction)  Homework 0 due  NO EXTENSION DAYS ALLOWED 
Tue, Aug 30  Lecture 3: nonlinear feature mappings, kernels (introduction), kernel perceptron  Homework 0 solution 
Thu, Sep 1  Lecture 4: SVM with kernels, dual solution  Homework 1: due on Sep 8, 11.59pm EST 
Tue, Sep 6 
Lecture 5: oneclass problems (anomaly detection), oneclass SVM, multiway classification, direct multiclass SVM Refs: [1] [2] [3] [4] (not mandatory to be read) 

Thu, Sep 8 
Lecture 6: rating (ordinal regression), PRank, ranking, rank SVM Refs: [1] (not mandatory to be read) 
Homework 1 due 
Tue, Sep 13  Lecture 7: linear and kernel regression, feature selection (information ranking, regularization, subset selection)  
Thu, Sep 15  Lecture 8: ensembles and boosting  Homework 2: due on Sep 27, 11.59pm EST 
Tue, Sep 20  —  
Thu, Sep 22  —  
Tue, Sep 27 
Lecture 9: model selection (finite hypothesis class) Refs: [1] (not mandatory to be read) 
Homework 2 due 
Thu, Sep 29 
Lecture 10: model selection (growth function, VC dimension, PAC Bayesian bounds) Notes: [1] 
Project plan due (see Assignments for details) 
Tue, Oct 4 
Lecture 11: generative probabilistic modeling, maximum likelihood estimation, mixture models, EM algorithm (introduction) Notes: [1] 

Thu, Oct 6 
Lecture 12: mixture models, EM algorithm, convergence, model selection Notes: [1] 

Tue, Oct 11  OCTOBER BREAK  
Thu, Oct 13 
Lecture 13: active learning, kernel regression, Gaussian processes Refs: [1] (not mandatory to be read) 

Tue, Oct 18  MIDTERM  3.00pm4.15pm at SengLiang Wang Hall 2599 
Thu, Oct 20  (midterm solution)  
Tue, Oct 25 
Lecture 14: collaborative filtering (matrix factorization), structured prediction (maxmargin approach) Refs: [1] (not mandatory to be read) 

Thu, Oct 27  (lecture continues)  
Tue, Nov 1  Lecture 15: performance measures, crossvalidation, biasvariance tradeoff, statistical hypothesis testing  Preliminary project report due (see Assignments for details) 
Thu, Nov 3  —  
Tue, Nov 8  Lecture 16: dimensionality reduction, principal component analysis (PCA), kernel PCA  
Thu, Nov 10  —  
Tue, Nov 15 
Lecture 17: Bayesian networks (motivation, examples, graph, independence) Refs: [1] [2] (not mandatory to be read) 

Thu, Nov 17 
Lecture 18: Bayesian networks (independence, equivalence, learning) Refs: [1] [2] [3, chapters 1620] (not mandatory to be read) 
Homework 3: due on Nov 22, 11.59pm EST 
Tue, Nov 22  —  Homework 3 due 
Thu, Nov 24  THANKSGIVING VACATION  
Tue, Nov 29 
Lecture 19: Bayesian networks (introduction to inference), Markov random fields, factor graphs Refs: [1] [2] (not mandatory to be read) 

Thu, Dec 1 
Lecture 20: Markov random fields (inference, learning) Refs: [1] [2] [3, chapters 1620] (not mandatory to be read) 
Final project report due (see Assignments for details) 
Tue, Dec 6  Lecture 21: Markov random fields (inference in general graphs, junction trees)  Not mandatory, extra Homework 4 posted on Kaggle 
Thu, Dec 8  —  
Wed, Dec 14  FINAL EXAM  8.00am9.30am at PHYS 223 