E. Mernyi
COMP / ELEC / STAT 502
3/18/14
page 1/2
Homework 8
Maximum available points: 100, 80 points = 100 %
Top three classifications of P1.b get 1 pizza point each, top three visualizations in P2, and
top three clustering in P2 get 2 pizza points each
E. Mernyi
COMP / ELEC / STAT 502
2/28/14
page 1/2
Homework 7
COMP / ELEC / STAT 502
Maximum available points: 75, 100% = 50 points
Problem 1. (20 points)
Generate a set of uniformly distributed 2-dimensional random vectors. The following MATLAB
command wi
E. Mernyi
COMP/ ELEC / STAT 502
2/19/14
page 1/1
Homework 6
ELEC / COMP / STAT 502
Problem 1. (50 points)
1.1
Calculate the principal eigenvectors (all four) of the iris data set, using the MATLAB routine
princomp (or equivalent). Use the first four colum
E. Mernyi
COMP / ELEC / STAT 502
page 1
Homework 5
The individuals achieving the 5 best results (on test data!) in the following categories will receive pizza
points J (one pizza point for each placement.):
5 best fits in Problem 1.a, 5 best fits in Probl
E. Mernyi
COMP / ELEC / STAT 502
Homework 4
page 1/2
Homework 4
The goal of these problems is to explore BP training, achieve good network performance: good
accuracy, faster convergence, or good accuracy with smaller training set. The individuals achievin
E. Mernyi
COMP / ELEC / STAT 502
Homework 3
1/30/2014
page 1/3
Homework 3
The goal of these two problems is to get everyone implement a BP network correctly, and get a feel for
network training, therefore the training / testing accuracy achieved will not
E. Mernyi
e
COMP / ELEC / STAT 502
Homework 2
1
Homework 2
Total possible score: 80 points (80 points = 100%)
Problem 1. (10 points total)
1/a (3 points)
If f (z) is the logistic (sigmoid) function
f (z) =
1
1 + eaz
(1)
where a is the slope parameter, sho