6.867 Machine Learning

6.867 Machine Learning

Fall 2002

Lectures

Lectures are Tue/Thu 2:30-4pm in E37-212

Lecture information and slides:

  • Lecture 1 (September 5) Introduction
    pdf slides or 6 per page (postscript)

  • Lecture 2 (September 10) Linear regression
    (Jordan & Bishop: chapters 5-5.3)
    pdf slides or 6 per page (postscript)

  • Lecture 3 (September 12) Additive models, maximum likelihood
    (Jordan & Bishop: chapter 4 (up to eq 4.20); chapters 5-5.3 & 5.6)
    pdf slides or 6 per page (postscript)

  • Lecture 4 (September 17) Active learning
    pdf slides or 6 per page (postscript)

  • Lecture 5 (September 19) Classification
    (Jordan & Bishop: chapter 4 mixture models; chapter 6-6.3.1)
    pdf slides or 6 per page (postscript)

  • Lecture 6 (September 24) Logistic regression, regularization
    pdf slides or 6 per page (postscript)
    Additional notes on regularization (pdf file)

  • Lecture 7 (September 26) Regularization, Support vector machines
    pdf slides or 6 per page (postscript)
    Tutorial on Lagrange multipliers (postscript)
    Optional paper on support vector machines (postscript)

  • Lecture 8 (October 1) Support vector machines, text classification
    pdf slides or 6 per page (postscript)

  • Lecture 9 (October 3) Feature selection, combination of methods
    pdf slides or 6 per page (postscript)
    Optional papers: boosting tutorial (postscript) and statistical view of boosting (postscript)

  • Lecture 10 (October 8) Boosting, complexity
    pdf slides or 6 per page (postscript)

  • Lecture 11 (October 10) Structural risk minimization, description length (+ midterm discussion)
    pdf slides or 6 per page (postscript)

  • MIDTERM (October 17) (in class)

  • Lecture 12 (October 22) Mixture models, EM
    (Jordan & Bishop: chapter 4 statistical concepts, mixture models; chapter 9 up to 9.2)
    pdf slides or 6 per page (postscript)

  • Lecture 13 (October 24) EM, regularization, conditional mixtures
    (Jordan & Bishop: chapter 9 up to 9.2.3)
    pdf slides or 6 per page (postscript)

  • Lecture 14 (October 29) Non-parametric density estimation, clustering
    pdf slides or 6 per page (postscript)

  • Lecture 15 (October 31) Clustering, Markov models
    pdf slides or 6 per page (postscript)

  • Lecture 16 (November 5) Markov and hidden Markov models
    (Jordan & Bishop: chapter 11)
    pdf slides or 6 per page (postscript)
    Optional tutorial on HMM (and some applications in speech recognition) (pdf)

  • Lecture 17 (November 7) Hidden Markov models
    pdf slides or 6 per page (postscript)

  • Lecture 18 (November 12) Viterbi, graphical models
    pdf slides or 6 per page (postscript)

  • Lecture 19 (November 14)Bayesian networks
    pdf slides or 6 per page (postscript)

  • Lecture 20 (November 19)Medical diagnosis example, influence diagrams
    pdf slides or 6 per page (postscript)

  • Lecture 21 (November 21)Influence diagrams, exact inference
    pdf slides or 6 per page (postscript)

  • Lecture 22 (November 26) Belief propagation
    pdf slides or 6 per page (postscript)

  • Lecture 23 (December 3) Learning graphical models (guest lecture)
    pdf slides or 6 per page (postscript)

  • FINAL (December 5) (in class)

  • Lecture 24 (December 10) no lecture, check email