| Date |
Lecture |
Notes etc |
| Wed. January 18th |
Lecture 1: Introduction, hidden Markov models, the perceptron
|
Slides:
Introduction,
HMMs,
the perceptron.
Background reading:
Notes on HMMs from COMS 4705, Fall 2011.
In addition,
Jurafsky and Martin Chapter 5 gives useful background on part-of-speech tagging.
|
| Wed. January 25th |
Lecture 2: SVMs, Pegasos, Log-linear models (part 1)
|
Slides on SVMs,
Slides on Pegasos,
Slides on log-linear models (part 1)
Background reading:
note on convergence proof for the perceptron
|
| Wed. February 1st |
Lecture 3: Log-linear models, MEMMs, CRFs
|
Slides:
Slides
Required reading:
Note on log-linear models, MEMMs, CRFs
Background reading (optional; gives additional detail on log-linear models):
Notes on Log-linear models from COMS 4705, Fall 2011.
|
| Wed. February 8th |
Lecture 4: CRFs (continued), discriminative
dependency parsing
|
Slides:
Slides
Required reading:
Note on the forward-backward algorithm
|
| Wed. February 15th |
Lecture 5: Dependency parsing (continued),
the structured perceptron, structured Pegasos, discriminative context-free
parsing
|
Slides:
Structured perceptron,
discriminative CFGs
(See last lecture slides for dependency parsing. Notes on structured
Pegasos to be posted soon.)
Background reading (optional; gives additional detail on context-free grammars):
Notes on probabilistics context-free grammars from COMS 4705, Fall 2011.
|
| Wed. February 22nd |
Lecture 6:
Naive Bayes Model, the EM Algorithm
|
Slides:
Slides part 1,
Slides part 2.
(Note that Gaussian mixture models (GMMs), and EM for GMMs, will
be covered on Feb. 29th)
|
| Wed. February 29th |
Lecture 7: The EM Algorithm (continued),
Gaussian mixture models, EM for HMMs
|
Slides:
EM for HMMs.
(See last lecture slides for slides on GMMs, and EM for GMMs)
|
| Wed. March 7th |
Lecture 8:
The Inside Outside Algorithm
|
Required reading:
Note on the EM algorithm,
Note on the inside-outside algorithm.
|
| Wed. March 14th |
Spring Break
|
|
| Wed. March 21st |
Lecture 9: Dual decomposition and
Lagrangian Relaxation
|
Slides (we went
as far as the slide titled
"Integrated Constituency and Dependency Parsing: Accuracy").
|
| Wed. March 28th |
Lecture 10: Dual decomposition and
Lagrangian Relaxation (continued)
|
Slides on non-projective dependency parsing,
Slides on phrase-based translation
|
| Wed. April 4th |
Lecture 11: Phrase-based Translation (continued),
and Cotraining
|
For slides on phrase-based translation see the last lecture.
Slides on cotraining.
|
| Wed. April 11th |
Lecture 12:
|
|
| Wed. April 18th |
Lecture 13: Final Exam
|
|
| Wed. April 25th |
Lecture 14:
|
|