Class: Backo (sections 4.6 and 4.7)
October 1, 2012
Next week Adi will talk on varaible length markov chains, and
Jordan will talk on HMMs.
Ill be setting up both of their talks this week.
Good prediction means good compression
means good mo
Homework 1: Regular expressions (due Sept 24 at
1. Read chapters 1 and 2 from JM.
2. From the book JM: 2.1, 2.4, 2.8
3. Exploritory data analysis is a common thing to do with numbers, histograms, box plots, etc. But, much of this isnt that inter
Class: Statistical parsing
November 12, 2012
Todays material is from Parameter Estimation for Statistical
Parsing Models: Theory and Practice of Distribution-Free Methods by Michael Collins.
Context free grammar epitomizes
Class: Speech encoding (chapter 7 & 8)
September 24, 2012
I posted homework 2. Due it two weeks. On n-grams.
Suppose I tell someone my home phone number while talking to them
on my cell phone:
The thoughts are digital
Class: Speech decoding (chapter 9 & 10)
September 26, 2012
How hard a problem to solve?
Trival: recognize Rex (1920)
Easy: recognize yes vs no
harder: recognize 0-10.
very hard: recognize English with careful speaker.
RIC: Risk Ination Criterion
October 24, 2012
Sorry about being slow on grading homework
Ive been moving to NYCso behind on lots of stu
Game plan for homework 4
Problem of domain adaption. Examples:
train on NYT to test on Wa
Class: Regular expressions
September 28, 2010
Regular expressions: key idea
rm *.* We all know what it means
regular expressions are a generalization of this
Idea is to capture a pattern
Basic tool in most string handling. Like a vector in math. Ev-
Class: Parsing (chapter 13)
November 7, 2012
Implement a streaming algorithm for HM 4.
Please read chapter 13 and 14 for next time
What is the advantage of generating a parse tree over simple POS?
POS is easy HMM and fast (it is
Class: Ngrams (chapter 4)
September 12, 2012
questions on homework?
Python class for statistiican tomorow at noon (by Josh and Justin,
I posted two papers to read before mondays class: JL and Fast
matrixes. We will revisit thes
September 17, 2012
Wednesday, Lyle will talk on Eigenwords
Next monday (Sept 24) we will move to R2D2 441. (Right hand
glass conference room in statistics.)
Draw some pictures
Draw d/2 pairs of axis
Draw d/2 vector
September 5, 2012
Who are you? Please answer the following questions
Name, program, degree
Background in stat
Background in lingustics
What did you hear this class was about? (Since it wasnt
announced, Im curious wha
December 3, 2012
Bootstrapping from truth
(Recall Bootstrap was a term before it was taken up by statistics. The
original usage mean, pulling yourself up by your bootstraps. It really
has little to do with Efrons ideabut
Class: Context free (chapter 12 & 16)
October 14, 2010
Another homework for your joy and pleasure
A, B, C: rules / states
a, b, c: terminals (actual words / symbols)
, , : mixture of the above
A Statistical MT Tutorial Workbook
prepared in connection with the JHU summer workshop
April 30, 1999
The Overall Plan
We want to automatically analyze existing human sentence translations, with an eye toward building
general translation ru