MAB608 Machine Learning


Overview Schedule and Reading Resources

Schedule


Organizational Meeting

(August 16) Overview of machine learning [slides odp|slides pdf] [notes1.txt]

Linear Regression and Review of Linear Algebra and Probability

(August 18) Curve Fitting and Least Square Error [homework1.pdf|input] [slides odp|slides pdf]

(August 23) Parameter Estimation I (Maximum Likelihood and Maximum Posterior) [slides odp|slides pdf]

(August 25) Parameter Estimation II (Fully Bayesian) and Decision Theory I [slides odp|slides pdf]
          [notes-decisiontheory.pdf]

Information Theory

(August 30) Decision Theory II and Information Theory I
          Due September 13 [homework2.pdf|input2] [slides odp|slides pdf] [notes5.txt]

(September 1) Information Theory II (Applications) [paper about image alignment|slides pdf]

Unsupervised Learning


Markov Chains and Google PageRank

(September 6) Random Walks and Google PageRank I (Intro to Markov Chains) [slides 1 pdf] [notes7.txt]

(September 8) Random Walks and Google PageRank II (Steady State Solution)
          [slides 2 pdf] [notes8.txt|example1.m]

(September 13) Random Walks and Google PageRank III (Transient Solution) [notes9.txt|example2.m]

(September 15) Random Walks and Google Page Rank IV (Continous Time I)
          [notes10.txt] [BrowseRank paper|Uniformization paper]

(September 20) Random Walks and Google Page Rank V (Continous Time II) [notes11.txt]

Statistical Models

Classical Models

(September 22) Statistical Models I [slides ppt|slides pdf] [notes12.txt]

(September 27) Statistical Models II [likelihood2.pdf] [homework3.pdf|data3] Due October 13

(September 29) Statistical Models III, Stochastic Approximation and Matlab [slides pdf|slides ppt] [Bonus homework on barycentric coordinates (Optional)]



(October 4) No Class Today (JIC)

(October 6) No Class Today (JIC)


(October 11) No Class Today (GBR)

Graphical Models

(October 13) Bayesian Networks I: Graphical Models [slides pdf]

(October 18) Bayesian Networks II: Graphical Models - D-Separation [slides pdf|alice.pdf|Mac Kay's book (chapter 16)]

(October 20) Bayesian Networks III: Graphical Models - Markov Random Fields, Inference [slides pdf|chater14a.pdf (Norvig and Russel)|chapter14b.pdf (Norvig and Russel)|see also chapter 16 of McKay's book]

Further reading: Detecting Network Neutrality Violations with Causal Inference, M. Tariq, M. Motiwala, N. Feamster, M. Ammar

To compute the confidence interval of measures obtained using sampling, see section 16.7.6 of Performance Evaluation material (in Portuguese)

Homework 4 [input file: original file with spaces original file corrupted file corrupted file with spaces] (Question 1 due October 27, Question 2 due November 1)

Supervised Learning

Linear Classification

(October 25) Linear Classification I [slides pdf|slides ppt|slides odp]
(October 27) Linear Classification II (Least Square, Fisher Discriminant and the Perceptron) [slides pdf|slides odp]

Non Linear Classification

(November 1) Neural Networks (and Support Vector Machines) [slides pdf] [slides pdf|slides odp] [perceptron.py|TODO]

Additional resources:

Training Neural Networks with Genetic Algorithms [Smart Sweepers|smart_sweepers]

Matlab Neural Network Toolbox [nnet_ug.pdf]

Unsupervised Learning


Clustering

(November 3) Expectation Maximization and K-Means [slides pdf|slides odp]

An article on spectral clustering: [On Spectral Clustering: Analysis and an algorithm, Ng, Jordan, Weiss, NIPS, 2002]

(November 8) Guest lecture by Prof Adilson Xavier on Clustering by Hiperbolic Smoothing

Learning To Act


Sequential Models

(November 10) Hidden Markov Models I (Inference) [A Tutorial on Hidden Markov Models, Rabiner|chapter15.pdf|bayesian networks taxonomy]



Homework 5 [input file: seq1.txt|seq2.txt] Due November 29

Reinforcement Learning

(November 17) Hidden Markov Models II (Viterbi and EM) and Markov Decision Processes I [mdp.gif]

(November 22) Markov Decision Processes II: Markov Processes with Rewards [slides pdf, mdp09.pdf]

(November 24) Markov Decision Processes III: Bellman Equations and Linear Programming [slides pdf, mdp09.pdf|article on MDPs and Stochastic Games]

(if time allows, POMDP, Reinforcement Learning and Stochastic Games)


Data Mining

(November 29) Weka and Decision Trees [decision trees|slides by F. Firmino|Google refine]




Further topics: Sampling

Markov Chain Monte Carlo