c 1100AMT305122301

RegNo.:
APJ ABDUL ICALAM TECHNOLOGICAL \*,,:
Fiftlr Semester B.Tech Degree Regular and Supplementary Examination Deiernber- ?V\,Qmg Scheme)

Course Code: AMT 305

Course Name: INTRODUCTION TO MACHINE LEARNING

Max. Marks: 100 Duration:3 Hours

PART A
Marks
(Answer all questions; each question carries 3 marks)

I Define VC dimension and shattering of a dataset. (3)

2 What is meant by most specific and most general hypothesis? (3)

3 Is Logistic regression a classification model? Justiff your answer. (3)

4 Formulate Information gain and entropy used in decision tree classifiers. (3)

5 Calculate F measure from the following table. Consider Healthy as positive class.
Actual
Classification
Deceased leaf Healthy leaf (3)

Prediction Healthy leaf 45 l8

Deceased leaf l2 25

6 What is the need of validation data while training a model? (3)

7 Conceptualise Maximum A-priori Estimation. (3)

8 Derive the equation for margin between support vectors in SVM? (3)

9 Point out the difference between voting and bagging (3)

l0 Explain the terms used to define high density areas in density based clustering (3)

algorithms? r
pfhr n
(Answer onefull questionfrom each module, each question carries 14 marks)

11 a). Define machine learning: *4rr ,rI;Jtlt"l<1,"0,., describe tr," uariois machine (14)
' learning paradigms.

12 a) Define the terms model selection and generalisation (8)

b) Show that an Axis aligned rectangle can shaffer only 4 point in two dimension. (6)

Page 1 of 3
 1100AMT30512230L

Module -2
13 a) Describe the feature selectioh method of dimensionality reduction. (6)

b) Consider the below frequency table which shows the frequency of each of the
attributes. Apply NaiVe Bayes algorithm to predict the fruit type of {Yellow,
Sweet, Long).

Fruit Yellow Sweet Long Total

Mango 350 450 0 650 (8)

Banana 400 300 350 400

Other fruits 50 100 50 150

Total 800 850 400 1200

14 a) Describe the different regression models and formulate any one enor (9)

measurrements used in regression analysis.

b) State ID3 algorithm, used for decision tree classification. (5)

Module -3
15 a) Define Sensitivity, Specificity, Accuracy and F- measure. Estimate these measures (10)
from the following table. The table describes the output of some classification
algorithm. Consider'Rainy Day' as positive class.
Sl. No Actual Value Predicted Value

I Rainy Day Rainy Day

2 Sunny Day Sunny Day

3 Sunny Day Sunny Day

4 Rainy Day Rainy Day

) Sunny Day Sunny Day

6 Sunny Day Sunny Day

7 Rainy Day Sunny Day

8 Sunny Day Rainy Day

g' Sunny Day Sunny Day

l0 Rainy Day Sunny Day

b) What is the significance of ROC curve in machine learning? (4)

16 a) Describe the working of multilayer perceptron. (8)

Page 2 of 3
 1100AMT305122301

b) Draw the perceptron for the following data and calculate the output of the (6)
perceptron if the following activitions are used?

Xl=l Wl:0.7
x24 W2:(_0.3)
X3:l W3=0.4
X4=l 1674:1-0.8)

X5:0 W5=0.5
a) Sigmoid activation fuction
b) TanH activation function
.
Module -4
17 a) Suppose the dataset (xl,x2,...xn) are independent and identically distributed (9)
drawn from exponential distribution exp(i). Find the maximum likelihood of t
b) Describe the Bias-variance decomposition. (5)
l8 a) Describe soft-margin SVM with neat diagram. (g)
b) Define any 3 kernal functions used for non-linear SVM (6)
Module -5
19 a) Use K-means clustering algorithm for the following data of heights and weights. (S)
Choose the first two data points as initial cluster centres. Use Euclidean distance

measune and identiff the cluster centre in the second iteration.

Height 185 170 168 179 r82 188 180 180 183 180 r80 177
Weight 72 56 60 68 72 77 7l 70 84 88 67 76
b) Illustrate the terms (6)
a) Single linkageb) Complete linkage c)Centroid linkage
' 20 a) Describe Density based clustering in detail. (8)
b) State any 3 similarity measures used in clusteriTq.algorithms (6)'
{.rt*

Page 3 of 3