Page 1 of 3
Name:
Reg No:
APJ ABDUL KALAM TECHNOLOGICAL UNIVERSITY
07 THRISSUR CLUSTER
SECOND SEMESTER M.TECH. DEGREE EXAMINATION APRIL 2018
DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING
COMPUTER SCIENCE AND ENGINEERING
07-CS-6106 MACHINE LEARNING AND LANGUAGE PROCESSING
Time:3 hours Max.marks: 60
Answer all six questions. Part of each question is compulsory.
Answer either part or part of eachquestion
Q.no. Module 1 Marks
1a Describe the working of Naïve Byes classifier. 4
Answer b or c
b Consider the following training data which is used for classification using decision tree.
This data will be analysed and classified in to two classes, One is buys_ computer
other is not buys_computer. What is Information gain? Find Information gain for the
attributes age.
Rid Age Class: Buys_
Computer
1 Youth No
2 Youth No
3 Middle-aged Yes
4 Senior Yes
5 Senior Yes
6 Senior No
7 Middle-aged Yes
8 Youth No
9 Youth Yes
10 Senior Yes
5
c Discuss in detail about Support Vector Machines and its applications. 5
Page 2 of 3
Q.no. Module 2 Marks
2a What is Logistic regression? How we can do classification of data using this
method?
4
Answer b or c
b Suppose you are a marketing analyst for Disney toys. You gather the following data.
After determining, via a scatter-plot, that the data followed a linear pattern, the
regression line was found. What is the relationship between sales and
advertisement?
Advertisement(x) Sales(y)
2 3
4 7
6 5
8 10
5
c Consider a scenario of online shopping website which sells different kind of
magazines. As a review manager, you have to collect all reviews and needs to label
them. Describe the idea of multi-label classification methods with this example.
5
Q.no. Module 3 Marks
3a Illustrate the idea of Expectation-Maximization algorithm. 4
Answer b or c
b With the help of an example explain about K-means clustering in detail. 5
c What is a sequential classifier? Write down the basic ideas of HMM and its
applications in detail with necessary examples.
5
Q.no. Module 4 Marks
4a What are Maximum Entropy Models and its significance? 4
Answer b or c
b Consider we had three urns urn urn 2 and urn these are three containers, and
they have distributions of red, green and blue balls within them.100 balls are in total
for each urn. Number of balls in each urn is shown below.
BALLS URN1(U1) URN2 URN3
RED(R) 30 10 60
GREEN(G) 50 40 10
BLUE(B) 20 50 30
Find Transition probability matrix and observation likelihoods. Also draw the Viterbi
tree for the above data.
5
Page 3 of 3
c Write a short note on Hidden Markov Model and three problems. How the
basic three problems of HMM can be solved?
5
Q.no. Module 5 Marks
5a Explain in detail about POS tagging and its applications.. 5
Answer b or c
b Find out the relative frequency bigram probabilities) of the following corpus
I want to study Chinese
I 5 827 0 9 0
want 2 0 608 1 6
to 2 0 4 686 2
study 0 0 2 0 16
Chinese 1 0 0 0 0
for 15 0 15 0 1
an 2 0 0 0 0
interview 1 0 1 0 0
Consider the following vocabulary size 1416and Count of each word in the
corpus is
I want to study Chinese
2533 927 2417 746 158
7
c
What is Named Entity Recognition How NER can be created using statistical
sequence labelling approach and just mention about evaluation measures of NER.
7
Q.no. Module 6 Marks
6a Explain about Factoid Question answering. How it is important for a search engine? 5
Answer b or c
b Discuss in detail about the strategies used for machine translation. Explain with an
example.
7
c Discuss in detail about the architecture of a speech recognizer and some
applications of Automatic Speech Recognition.
7