Essay Available:

Pages:

0 pages/≈0 words

Sources:

Check Instructions

Style:

Check Instructions

Subject:

IT & Computer Science

Type:

Exam

Language:

English (U.S.)

Document:

MS Word

Date:

Total cost:

$ 5

Topic:

# Machine Learning

Instructions:

Text classification is a core machine learning technique that can help you label your texts. Text analysis (TA) is a machine learning technique used to automatically extract valuable insights from unstructured text data.

Content:

Test Pearl 100 Intelligent Interaction
October 2 2015
The test consists of 5 questions. The grade is the number of achieved points divided 100.
1 (20 points) A bagH
1 contains
10marbles: 2 red, 3 white and 5 blue. Bag H
2 contains also
10marbles:
+
4 red, 2 white and 4 blue. Someone throws a fair dice and if the outcome is divisible by 3 then he chooses
bag H
1, else he chooses bag
H
2. After choosing a bag he draws 5 marbles with replacement. The outcome
D is 2 red, 2 white and 1 blue marble, so D=< 2;2 ;1 > .
(a)Compute P(D jH
1)
.
(b)What is the most likely bag from which the marbles are drawn; H
1or
H
2? Motivate your answer by a
computation using Bayes law. Antwoord op 1.
1.P(D jH
1) = 5! 2!1!2!
(2 10
)2
( 3 10
)2
( 5 10
) =
0:054
2. P(D jH
2) = 5! 2!1!2!
(4 10
)2
( 2 10
)2
( 4 10
) =
0:0768. And P(H
1j
D )= P (H
2j
D ) equals
[( 3 10
)2
( 5 10
)( 1 3
)]
=[( 4 10
)2
( 4 10
)( 2 3
)] =
45=128 =0:35. Hence H
2is the most likely bag.
2 (20 points) Given the following piece of text from an email:
+
attention if you are in debt. if you are then we can help. qualifying is now at your ngertips and
there are no long distance calls
(a)Assume that we use as vocabulary V=fattention, adult, debt, publications, qualifying, xxx g. How
would this piece of text be coded using a binary coding and this vocabulary V?
(b)For convenience consider a smaller vocabulary V=fattention, adult, debt gand assume that we have
a dataset consisting of 100 emails of which 30 are spam and with the following vocabulary frequency
list: Word Ham Spam
attention 30 10
adult 0 22
debt 4 20
This means for instance that the word attention occurs in 30 ham emails and in 10 spam emails.
Assume that a new email arrives with binary coding <1>. Compute the likelihood that this
email is from the spam class. in other words compute P(< 1;0 ;1 > jSpam ).
(c)How i

Get the Whole Paper!

Do you need a custom essay? Order right now: