 No credit for late homework.
 Programming assignments must be in Python, and must run on your instructional account on cycle1.csug.rochester.edu.
 Use numpy for matrix operations.
 To get an instructional account, see Brynn Wilkins in WH 2513.
 To turn in programming assignments, run the command /u/cs246/TURN_IN
 Homework 1, due in class Mon 1/22
 Bishop 1.3
 Bishop 1.11
 Let (X ⫫ Y) denote that X and Y are independent, and let (X ⫫ YZ) denote that X and Y are independent conditioned on Z (Bishop p. 373).
Are the following properties true? Prove or disprove.

(X ⫫ WZ,Y) ∧ (X ⫫ YZ) ⇒ (X ⫫ Y,WZ)

(X ⫫ YZ) ∧ (X ⫫ YW) ⇒ (X ⫫ YZ,W)
 Homework 2, due Fri 1/26 5pm by turnin script

Implement perceptron
for the adult income dataset using Python.
Data available in /u/cs246/data/adult.
Experiment with performance as a function of number of iterations.
See detailed interface instructions.
 Homework 3, due Fri 2/2 5pm by turnin script

Implement SGD for SVM
for the adult income dataset.
Experiment with performance as a function of the capacity parameter C.
See detailed interface instructions.
 Homework 4, due 2/14 in class

Bishop 4.8, 5.4, 5.7, 5.17
 Practice Problems: not collected, finish by 2/23

Bishop 6.2, 6.11 (we did this informally in class, check that you can do it formally), 7.2, 7.4
 Homework 5, due 2/28 in class

Review problems handed out in class.
 Homework 7, due Fri 3/30 5pm by turnin script

Implement EM fitting of a mixture of gaussians on the twodimensional
data set points.dat. You should
try different numbers of mixtures, as well as tied vs. separate
covariance matrices for each gaussian.
OR
Implement EM fitting of the aspect model on the
discrete data set pairs.dat.
You should try different numbers of mixtures.
IN EITHER CASE
Use the final 1/10 of the data for dev.
Plot likelihood on train and dev vs iteration
for different numbers of mixtures.
 Homework 8, Fri 4/20 5pm by turnin script

Implement EM to train an HMM for whichever dataset
you used for assignment 7.
The observation probs should be as in assignment 7:
either gaussian, or two discrete distributions conditionally
independent given the hidden state.
Does the HMM model the data better than the original
nonsequence model? What is the best number of states?
 Practice Problems: not collected

Bishop 9.8, 9.9, 11.12, 11.13
 Homework 9, Mon 4/30 in class

Review problems handed out in class.
gildea @ cs rochester edu
January 31, 2018
