Computational Theories of Learning and Reasoning

Fall 1999

Lecture Notes

Lectures notes will be available here as gzipped postscripts and source (latex) files.
The style file for scribing the lectures is here.
Most notes are based on a draft from the 97 lectures. Ask for the draft before you start working on your lecture notes.

  1. Lecture #1:
             Introduction to Computational Learning;The PAC model
    (8/31,9/2) tex
  2. Lecture #2:
             The importance of representation: Learning 2-term DNF is hard.
    (9/7,9/9) tex
  3. Lecture #3:
             Occam's Razor: VC dimension.
    (9/16,9/18) tex
  4. Presentation #1:
             Margin Distribution Bounds on Generalization (Presented by Ming-Hsuan Yang) tex
  5. (9/21,9/23)
  6. Lecture #4:
             On Line Learning
    (9/28,9/30) tex
  7. Presentation #2:
             Projection Learning (Presented by Shripad Thite)(10/7) tex
  8. Presentation #3:
             The Robustness of p-norm algorithms (Presented by Shaojun Wang)
    (10/12,10/14)
  9. Presentation #4:
             Drifting Games (Presented by Ting-Hao Yang)
    (10/19)
  10. Lecture #5:
             Active Learning
    (10/21,10/26) tex
  11. Presentation #6
             : Learning Automata (Presented by Vasin Punyakanok)
    (10/28)
  12. Presentation #5:
             Combining labeled and unlabeled data with Co-Training (Presented by Dav Zimak)
    (11/2)
  13. Lecture #6:
             Reasoning with Models
    (11/9,11/4) tex
  14. Lecture #7:
             Learning to Reason
    (11/16,11/11) tex
  15. Presentation #7:
             Learning to take Action (Presented by Oleg Pashko)
    (11/11)
  16. Lecture #8:
             Introduction to Probabilistic Reasoning
    (11/18) tex
  17. Presentation #8:
             Query DAGS: A paradigm for implementing Belief Network Inference (Presented by Sun Qiang )
    (11/18)
  18. Presentation #9:
             Correction of belief propagation in Gaussian graphical models (Presented by Cedric Yau)
    (11/23)


Dan Roth