Syllabus and Course Schedule

Module

Lectures, readings, online materials 

Homeworks, Exams

Supervised Learning



Intro to Functional Aproximation

 Lecture 1: 1/14/08

    Slides
    (annotated slides)

    Overview of Machine Learning

    Basic Probability Theiry (Recitations)

    Density Estimation

  • Maximum likelihood principle
  • Overfitting
  • Bayesian versus Frequentist estimate

Reading:
  • Biship book: Chapter 1, 2


Lecture 2: 1/16/08

    Classification Theory: Optimal Classifier

    Nonparametric methods & Instance-based learning

  • Bayesian decision rule
  • Bayes error
  • Parzen and nearest neignbor density estimation
  • K-nearest neighbor (kNN) classifier

Reading:

  • Bishop book: Chapter 2.5
  • Fukunaga (Intro to Statistical PR)

 

 















Approximating Linear Seperation Function
 Lecture --: 1/21/08
 

Celebrating MLK Day --- No class

 

 
HW1 out
(handout, data for Problem 5, solution)

Lecture 3: 1/23/08

Slides
(annotated slides)
Generative classifiers

  • Naive Bayes classifiers [Applet] with discrete  and continuous (Gaussian) features

Reading:


 

 

 Lecture 4: 1/28/08

Slides
(annotated slides)
Discriminative classifiers:

  • Linear regression [Applet] and its probabilistic interpretation

Reading:

  • Bishop: chap. 3
  • Mitchell: chap. 8.3

 

 

 

Lecture 5: 1/30/08

  • Logistic regression [Applet]
  • Relationship to Naive Bayes
  • Linear regression and its probabilistic interpretation

Reading:

 











Approximating Non-Linear Seperation Function

 Lecture 6: 2/4/08

Complex Discriminative Functional Learing:

Slides
(annotated slides)

 Decision Tree Learning [Applet]

Reading:

  • Mitchell, Chapter 3 on Decision Trees
  • Bishop, Section 1.6


 

HW1 due at the beginning of class.

 

Lecture 7: 2/6/08

Slides
(annotated slides)

Neural Network [Applet]

  • Non-linear regression, classifiers
  • Gradient descent
  • Discovered representations at hidden layers

Reading:

  • Bishop, Chap 5
  • Mitchell, Chap 4

HW2  out (handout,solution)  

 

 Lecture 8-9: 2/11-13/08

Slides
(annotated slides)

Support Vector Machine [Applet]

  • Duality
  • The Kernel methods
  • convex optimization

Reading:


 

 Lecture 10: 2/18/08

Slides
(annotated slides)

Boosting Weak Classifiers [Adaboost Applet]

  • Combination of clsasifiers
  • Ada boost
Reading:
the boosting homepage
 









Theory and Practice in Supervised Learning

Lecture 11: 2/20/08

Slides
(annotated slides)
Learning theory I:

  • Sample complexity
  • Hypothesis space
  • PAC learning theory [Applets]
  • Agnostic learning

 

HW2 due

HW3 out (handout)

Lecture 12: 2/25/08

Slides
(annotated slides)
Learning theory II:

  • VC dimension Agnostic learning
  • Overfitting and PAC bounds
  • Structural Risk Minimization


Project proposals due

Lecture 13: 2/27/08

Slides
(annotated slides)
Practical issues in supervised learning, and keys

  • Overfitting
  • Decomposition of error into bias and variance
  • Cross-validation
  • Regularization
Reading:
  • Bishop, Chap 1 & 2
  • Mitchell, Chap 5&6


Lecture 14: 3/3/08

Slides
(annotated slides)
Model selection, Feature selection

Reading:

 Review for the MidTerm

HW3 due


 Lecture (to be rearranged)

Application I: Text classification, spam-filtering

 

 





 

 MIDTERM

 


 

Midterm Exam (3/5/08): open book, open notes, no computers

 

 

MIDTERM (Solution)

 



Lecture --: 3/10/08

    No class (Spring break)

     

 


Lecture --: 3/12/08

No class (Spring break)

 


Unsupervised Learning and Structured Learning in High-Dimensional Space




Unsupervised Learning
Lecture 15: 3/17/08

Slides
(annotated slides)
Introduction to Unsupervised Learning

Reading:
  • Bishop, Chap 9

 

Lecture 16: 3/19/08

Reading:

  • Bishop, Chap 9

 







Learning Complex Function

 Lecture 17: 3/24/08

 Slides
(annotated slides)
Hidden Markov Models

  • Representation: discrete and continuous observations
  • Inference: the forward-backward algorithm
  • Learning: the Balm-Welsh (EM) algorithm
Reading:
  • Bishop, Chap 13


HW4 out

 Lecture 18: 3/26/08

Reading:

Project milestone due

 Lecture 19-20: 3/31-4/2/08

Reading:




 Lecture 21: 4/7/08 

  • How to decipher our DNA?: gene finding
  • Where are we from?: cncestral inference

Reading:

 

HW4 due

Understanding Graphs and Exploiting Linkage structures

Lecture 22: 4/9/08 

HW5 out


 Lecture 23: 4/14/08

Slides
(annotated slides)
Graph-Theoreric Methods for Clustering

  • Spectral clustering
  • normalized cut

Reading:

  • On Spectral Clustering: Analysis and an algorithm, Andrew Y. Ng, Michael Jordan, and Yair Weiss. In NIPS 14,, 2002. [ps, pdf]
  • Normalized Cuts and Image Segmentation, Jianbo Shi and Jitendra Malik, IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(8), 888-905, August 2000. [pdf]


Dimensionality Reduction


 Lecture 24: 4/16/08

Slides
(annotated slides)
Dimensionality reduction I:

  • Principal component Analysis [Applet]
  • Singular value decomposition
  • LSI
Reading:


Lecture 25: 4/21/08

Slides
(annotated slides)

  • Probablistic LSI
  • Hot topics: Topic models and natural language analysis
Reading:


Lecture 26: 4/23/08

Slides
(annotated slides)
Dimensionality reduction II:

  • Probabilistic PCA
  • Factor Analysis
  • Metric Learning
  • Independent Components Analysis

 
Reading:

  HW 5 due

Learning control strategies


Lecture 27: 4/28/08

Slides
(annotated slides)
Reinforcement learning I:

  • Markov decision processes
  • Learning control stategies when next-state function is known
  • Value iteration
  • Policy iteration

Reading:



 Lecture 28: 4/30/08 

    • Learning when next-state function is unknown
    • Q-Learning
    • Temporal difference learning in primates

Project final report due

 

5/1/08

    Project Poster presentation (NSH 3305, 3-6pm)

     

 

Final Exam

 5/6/08, 8:30am-11:30am (Scaife Hall (SH) 125)