CS 4803A/8803A: Pattern Recognition - Problem Set 1, Assignments of Computer Science

Problem set 1 for the cs 4803a/8803a: pattern recognition course. It includes various statistical independence problems, density function calculations, and a monty hall problem. Students are required to use matlab for some tasks.

Typology: Assignments

Pre 2010

Uploaded on 08/05/2009

koofers-user-kn7
koofers-user-kn7 🇺🇸

10 documents

1 / 2

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
CS 4803A/8803A: Pattern Recognition
Problem Set 1
Date: January 14, 2003 Due: start of class January 23, 2003
1. For each case below, are
and
statistically independent? Why or why not?
(a) The point


is uniformly distributed in the unit square.
(b) The point


is uniformly distributed in the unit circle (a subset of the unit square).
(c) Let
and
be independent and both uniform on


. Let


and

.
(d) Let
and
be independent and both Gaussian with zero mean and unit variance. Let


and

.
(e) Let
and
be independent and both Gaussian with zero mean, but the variance of
is 2
while the variance of
is 1. Again, let


and

.
2. Let
and
be random variables with density functions
! "
$#

&% ')(+*
(
"-,
/.
1032
10
and
"
4
%
5.36.
otherwise.
(a) What is
7"


? Specify the region where the joint density is nonzero.
(b) Using Matlab, generate a plot of
87"


.
(c) Find

and plot that as well. Don’t forget to specify regions of definition.
3. The components
9
and
;:
of a two-dimensional random vector
<
are statistically independent
and have the following marginal densities:
=
>9
%
5.
>9
.
otherwise
and
@?
8:
% A
8:
B.
8:
.
otherwise
C
Let
D
be the event
E9
.
8:
. What are the following?
(a)
F
4<G
(Plot it in 3D in Matlab.)
pf2

Partial preview of the text

Download CS 4803A/8803A: Pattern Recognition - Problem Set 1 and more Assignments Computer Science in PDF only on Docsity!

CS 4803A/8803A: Pattern Recognition

Problem Set 1 Date: January 14, 2003 Due: start of class January 23, 2003

  1. For each case below, are and  statistically independent? Why or why not?

(a) The point ^  is uniformly distributed in the unit square. (b) The point ^  is uniformly distributed in the unit circle (a subset of the unit square). (c) Let  and  be independent and both uniform on  ^. Let ^  and   . (d) Let  and  be independent and both Gaussian with zero mean and unit variance. Let  (^)  and   .

(e) Let  and  be independent and both Gaussian with zero mean, but the variance of  is 2 while the variance of  is 1. Again, let ^  and   .

  1. Let and  be random variables with density functions

! " $# (^)  &% ')(+*

/. ^1032

and  " (^) 4  % 5.36. otherwise.

(a) What is 7"^ ^ ? Specify the region where the joint density is nonzero. (b) Using Matlab, generate a plot of 87"^ ^ . (c) Find ^   and plot that as well. Don’t forget to specify regions of definition.

  1. The components 9^ and ;:^ of a two-dimensional random vector < are statistically independent and have the following marginal densities:

 =>9 (^)   % 5.^

>^.

otherwise

and  @? (^) 8: (^)   % A

8: B. 8: .

otherwiseC Let D be the event E9^. 8:^. What are the following?

(a) F^ 4<G (Plot it in 3D in Matlab.)

Write a Matlab simulation and verify your answer for the following: (b) HJI KD  (c) FL^ M N< #D (d)  ? M 8:O#^ DP

  1. The random variable has density

 (^)    % CRQOS^

S

otherwise

(a) What is the density of T VUXW^  ? (b) What is the density of  Y^ S ? What is the mode of 8"^ 4 , compared to the mode of  (^)  ?

(c) For discrete random variables, the mode of the probability distribution can be interpreted as the most likely value. Does this interpretation apply equally well to continuous random variables?

  1. A student takes a test to indicate whether or not the student is a nerd. The test is positive. Suppose the student population is known to contain 5% nerds (regardless of any test results). Suppose also that the nerd test is 75% accurate, with a false alarm rate of 20%. (In other words, if someone is a nerd, then the chance the test will come out positive is 75%; if they are not a nerd the chance of a positive result is 20%.) How probable is it that the student really is a nerd, given that the test is positive?
  2. Below is a variation on the classic “Monty Hall” problem. To get credit for this problem, you must pose it to at least one other person who have not understood it before, and help them find the correct answer. Have them sign your problem set with a statement that indicates you have succeeded in this goal. Problem: Suppose I hide the solution to the current problem set in one of three identical boxes while you aren’t looking. Then I ask you to guess which box it’s in. I know which one it’s in, and after you guess, I deliberately open the lid of an empty box, which is one of the boxes you have not chosen. Thus, the solutions are either in the box you guessed, or in the box (you didn’t pick) that I didn’t open. I then offer you the chance to change your mind: you can either keep the box you originally guessed, or choose the other unopened box. To maximize your chances of getting the problem set solution, what choice should you make? You must explain your answer.