Scribd Logo
Upload

Welcome to Scribd!

  • 20 views

    Machine Learning PYQ 2023

    Uploaded by

    nitob90303

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Machine Learning PYQ 2023

    Uploaded by

    nitob90303
    20 views8 pages

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    20 views8 pages

    Machine Learning PYQ 2023

    Uploaded by

    nitob90303

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 8

    [This question paper contains 8 printed pages.

    Your Roll NoZa».L3Slo oo 8


    Sr. No. of Question Paper : 4835 E

    Unique Paper Code : 32347607

    Name of the Paper : Machine Learning


    Name of the Course : B.Sc. (Hons.) Computer
    Science (LOCF)
    Year of Admission (Admission of 2019)
    Semester : VI

    Duration: 3Hours Maximum Marks : 75

    Instructions for Candidates


    1. Write your Roll No. on the top immediately on receipt
    of this question paper.

    2. Section A is compulsory.
    3. Attempt any four questions from Section B.
    4. Use of scientific calculator is allowed.

    Section A (Compulsory)

    1. (a) Explain the cost function of logistic regression.


    (3)

    P.T.0.
    4835 2

    (b) Give a formula for the binary sigmoid


    activation
    function. Also obtain the first derivative of this
    function. (3)
    (c) What is K-means clustering algorithm used
    for?
    (3)
    (d) Explain two applications of machine learning.
    (3)
    (e) Consider the following data. Find the information
    gain if A2 is selected as the root of the decision
    tree.
    (6)
    A1 A2 A3 Class
    True Hot High No
    True Hot High No
    Faise Hot High Yes.
    False Cool Normal Yes
    False Cool Normal Yes
    True Cool High No
    True Hot High No

    True Hot Normal Yes

    False Cool Normal Yes


    False Cool High Yes

    (t) Can a single layer perceptron solve the XOR


    problems? Justify your answer. (3)
    4835 3

    (g) Find the line of best fit using least square


    regression method, given the following data :

    X y
    2 6
    6 13
    10 28
    12 30

    Predict the value of y for the test data x = 11.


    (4)

    (h) Given two situations : (4)

    i) A publishing house wants to predict the


    number of copies of a book would be sold.
    They want to use the popularity rating of
    the author, success index of her/his previous
    books and one other variable.

    (ii) Based on the blood test reports, a


    researcher builds a model to identify the
    type of diabetes (type 1, type 2, gestational
    diabetes) that a patient has.

    Identify each of these situations as a classification


    problem or a regression problem. Justify your
    answer.

    P.T.0.
    4835 4

    () Consider the Confusion matrix given below:

    Actual Values
    True False
    Predicted True 10 11

    Values False 4 20

    Compute:
    (i) Accuracy

    (ii) Precision

    (ii) Recall (6)

    Section B

    2. (a) Differentiate between supervised, unsupervised and


    reinforcement learning. (6)

    (b) Using a diagram, depict the following scenarios


    for a machine learning model :

    (i)) High variance, lowbias

    (ii) Low variance, high bias (4)


    4835 5

    3. (a) It is found that a classification model performs


    with a high accuracy on training data, but with
    new instances, it generalizes poorly. What could
    be the problem in this model? Explain using a
    suitable diagram. Give two possible solutions to
    this problem. (6)

    (b) Define and elaborate the terms : (4)

    (i) Hypothesis space

    (ii) Inductive bias

    4. (a) Apply the K-nearest neighbor algorithm on the


    following dataset : (4)

    b Class
    7 7 False
    7 4 False
    4 True
    4 True

    Predict the class for X (a =3, b=7). Assume


    K=3 (Use Euclidean distance to compute the
    distance)
    P.T.0.
    4835 6

    (b) What is the disadvantage of the filter approach


    for fcature selection?

    List the steps for feature selection using Principal


    Component Analysis (PCA). (6)

    5. (a) Design a neural network for the Boolean function


    AND (for two variables) with the help of a neat
    diagram. (4)
    (b) Explain the back-propagation algorithm for
    multilayer perceptron. (6)

    6. (a) What do you understand by the following terms


    W.r.t. Support Vector Machine (SVM) learning
    algorithm?
    (i) Support vectors

    (i) Marginal distance (4)


    (b) For the following data, apply one
    iteration of
    K-means clustering to partition the data in 2
    clusters. Assume points C1 = (1,2,3) and
    C2 = (3.4.5) as the initial cluster centers.
    Show
    the generated clusters and cluster centers after
    the first iteration. (Use Manhattan distance to
    compute the distance). (6)
    4835 7

    A B C

    2 4 1
    2 3 1
    6
    5 2 9
    1 4

    7. (a) What is the basic assumption of a Naive Bayes


    classifier? (2)

    (b) Consider the following training data set for car


    theft :

    Sample No. Color Type Origin Stolen?


    Red Sports Domestic Yes
    2 Red Sports Domestic No .
    Red Sports Domestic Yes
    4 Yellow Sports Domestic No
    Yellow Sports Imported Yes .
    6 Yellow SUV Imported No
    Yellow SUV Imported Yes.
    Yellow SUV Domestic No
    9 Red SUV Imported No
    10 Red Sports Imported Yes

    Use the Naive Bayes classification rule to classify


    a Red Domestic SUV. (8)
    P.T.0.
    4835 8

    8. (a) Considerthe following linear regression problem :

    Xx 2 4 3 5
    1 3 3 2 5

    Given the initial values of ,, 6, as 0.5, 0.5 and


    learning rate as 0.1. Compute the next set of values
    for regression coefficients, using gradient descent
    method. (6)

    (b) Answer the following for 5-fold cross validation


    on atraining data set of 45 tuples :

    (i) How many rounds of learning will be


    performed?

    (ii) Statethe size of the training and testing set


    for each round of learning.

    (i1i) How would the accuracy of the classifier


    be calculated in 5-fold cross validation?
    (4)

    (500)

    You might also like