2022 question paper
MACHINE LEARNING
22 questions
Q1a. Which of the following is correct use of cross validation? (i) Selecting variables to include in a model (ii) Comparing predictors (iii) Electing parameters in prediction function (iv) All of the above20222m
Module 1: Introduction
View this question on its own page →Which of the following is correct use of cross validation?
(i) Selecting variables to include in a model
(ii) Comparing predictors
(iii) Electing parameters in prediction function
(iv) All of the aboveQ1b. Which of the following statement(s) is/are true for Gradient Decent (GD) and Stochastic Gradient Decent (SGD)? (1) In GD and SGD, you update a set of parameters in an iterative manner to function. minimize the error (2) In SGD, you have to run through all the samples in your training set for a single update of a parameter in each iteration. (3) In GD, you either use the entire data or a subset of training data to update a parameter in each iteration. (i) Only (1) (ii) (1) and (2) only (iii) (3) and (2) only (iv) (1), (2) and (3)20222m
Module 1: Introduction
View this question on its own page →Which of the following statement(s) is/are true for Gradient Decent (GD) and Stochastic Gradient Decent (SGD)?
(1) In GD and SGD, you update a set of parameters in an iterative manner to function. minimize the error
(2) In SGD, you have to run through all the samples in your training set for a single update of a parameter in each iteration.
(3) In GD, you either use the entire data or a subset of training data to update a parameter in each iteration.(i) Only (1)
(ii) (1) and (2) only
(iii) (3) and (2) only
(iv) (1), (2) and (3)Q1c. What makes the difference between good decisions and bad decisions? (i) A good decision is based on logic (ii) A good decision considers all available data (iii) A good decision considers all alternatives (iv) A good decision applies quantitative approaches20222m
Module 2: Linear Models and Bayesian Learning
View this question on its own page →What makes the difference between good decisions and bad decisions?
(i) A good decision is based on logic
(ii) A good decision considers all available data
(iii) A good decision considers all alternatives
(iv) A good decision applies quantitative approachesQ1d. Where does the Bayes rule can be used? (i) Solving queries (ii) Increasing complexity (iii) Decreasing complexity (iv) Answering probabilistic query20222m
Module 2: Linear Models and Bayesian Learning
View this question on its own page →Where does the Bayes rule can be used?
(i) Solving queries
(ii) Increasing complexity
(iii) Decreasing complexity
(iv) Answering probabilistic queryQ1e. Which of the following steps/assumptions in regression modeling impacts the trade-off between under-fitting and over-fitting the most? (i) The polynomial degree (ii) Whether we learn the weights by matrix inversion or gradient descent (iii) The use of a constant-term (iv) None of the above20222m
Module 2: Linear Models and Bayesian Learning
View this question on its own page →Which of the following steps/assumptions in regression modeling impacts the trade-off between under-fitting and over-fitting the most?
(i) The polynomial degree
(ii) Whether we learn the weights by matrix inversion or gradient descent
(iii) The use of a constant-term
(iv) None of the aboveQ1f. Which of the following is true about Naive Bayes? (i) Assumes that all the features in a dataset are equally important (ii) Assumes that all the features in a dataset are independent (ii) Both (i) and (ii) (iv) None of the above20222m
Module 3: Classification Algorithms
View this question on its own page →Which of the following is true about Naive Bayes?
(i) Assumes that all the features in a dataset are equally important
(ii) Assumes that all the features in a dataset are independent
(ii) Both (i) and (ii)
(iv) None of the aboveQ1g. The feature of ANN in which ANN creates its own organization or representation of information it receives during learning time is (i) adaptive learning (ii) self-organization (iii) what-if analysis (iv) supervised learning20222m
Module 3: Classification Algorithms
View this question on its own page →The feature of ANN in which ANN creates its own organization or representation of information it receives during learning time is
(i) adaptive learning
(ii) self-organization
(iii) what-if analysis
(iv) supervised learningQ1h. Which of the following is true about averaging ensemble? (i) It can only be used in classification problem (ii) It can only be used in regression problem (ii) It can be used in both classification as well as regression (iv) None of the above20222m
Module 3: Classification Algorithms
View this question on its own page →Which of the following is true about averaging ensemble?
(i) It can only be used in classification problem
(ii) It can only be used in regression problem
(ii) It can be used in both classification as well as regression
(iv) None of the aboveQ1i. What is the minimum number of variables/features required to perform clustering? (i) 0 (ii) 1 (iii) 2 (iv) 320222m
Module 4: Clustering, K-means, K-medoids, Density-based Hierarchical, Spectral .
View this question on its own page →What is the minimum number of variables/features required to perform clustering?
(i) 0
(ii) 1
(iii) 2
(iv) 3Q2a. Give the three computer applications for which machine learning approaches seem appropriate and three for which they seem inappropriate.20227m
Module 1: Introduction
View this question on its own page →Give the three computer applications for which machine learning approaches seem appropriate and three for which they seem inappropriate.
Q2b. What is decision tree algorithm? What are the issues in decision tree induction?20227m
Module 3: Classification Algorithms
View this question on its own page →What is decision tree algorithm? What are the issues in decision tree induction?
Q3a. Define how Bayesian networks and Markov random fields are used to represent probability distributions, and briefly describe the trade-offs involved in choosing one versus the other.20227m
Module 5: Probabilistic & Reinforcement Methods
View this question on its own page →Define how Bayesian networks and Markov random fields are used to represent probability distributions, and briefly describe the trade-offs involved in choosing one versus the other.
Q3b. What is regression analysis? What are the differences between linear regression and ridge regression?20227m
Module 2: Linear Models and Bayesian Learning
View this question on its own page →What is regression analysis? What are the differences between linear regression and ridge regression?
Q4. Explain how the linear maximum-margin classifier can be modified to be non-linear and to allow misclassification of the training examples. Give a derivation of the modified optimization problem needed for training. You need only derive a statement of the primal optimization problem.202214m
Module 3: Classification Algorithms
View this question on its own page →Explain how the linear maximum-margin classifier can be modified to be non-linear and to allow misclassification of the training examples. Give a derivation of the modified optimization problem needed for training. You need only derive a statement of the primal optimization problem.
Q5. Following is a data set that contains two attributes, X and Y, and two class labels, ‘+’ and ‘–’. Each attribute can take three different values: 0, 1, or 2. The concept for the ‘+’ class is Y = 1 and the concept for the ‘–’ class is X = 0 ∨ X = 2.  (a) Build a decision tree on the data set. Does the tree capture the ‘+’ and ‘–’ concepts? (b) What are the accuracy, precision, recall and F1-measure of the decision tree? (Note that precision, recall, and F1-measure are defined with respect to the ‘+’ class.)202214m
Module 3: Classification Algorithms
View this question on its own page →Following is a data set that contains two attributes, X and Y, and two class labels, ‘+’ and ‘–’.
Each attribute can take three different values: 0, 1, or 2.The concept for the ‘+’ class is Y = 1
and the concept for the ‘–’ class is X = 0 ∨ X = 2.
(a)
Build a decision tree on the data set.
Does the tree capture the ‘+’ and ‘–’ concepts?(b)
What are the accuracy, precision, recall and F1-measure of the decision tree?
(Note that precision, recall, and F1-measure are defined with respect to the ‘+’ class.)Q6a. Explain the model of over-fitting and under-fitting. Give the example of over-fitting and under-fitting in two-dimensional data.20227m
Module 1: Introduction
View this question on its own page →Explain the model of over-fitting and under-fitting. Give the example of over-fitting and under-fitting in two-dimensional data.
Q6b. Explain the bagging and boosting with example.20227m
Module 3: Classification Algorithms
View this question on its own page →Explain the bagging and boosting with example.
Q7. Discuss the steps in K-means algorithm and evaluate the following table using K-means : 202214m
Module 3: Classification Algorithms
View this question on its own page →Discuss the steps in K-means algorithm and evaluate the following table using K-means :
Q8a. What is the difference between K-means and K-medoid algorithm? Explain density-based hierarchical clustering.20227m
Module 4: Clustering, K-means, K-medoids, Density-based Hierarchical, Spectral .
View this question on its own page →What is the difference between K-means and K-medoid algorithm? Explain density-based hierarchical clustering.
Q8b. What is ANN? Explain feed forward and back propagation algorithm with example.20227m
Module 3: Classification Algorithms
View this question on its own page →What is ANN? Explain feed forward and back propagation algorithm with example.
Q9a. Write short notes on the following: (a) Partial least square20227m
Module 2: Linear Models and Bayesian Learning
View this question on its own page →Write short notes on the following:
(a) Partial least squareQ9b. Write short notes on the following: (b) Reinforcement learning20227m
Module 5: Probabilistic & Reinforcement Methods
View this question on its own page →Write short notes on the following:
(b) Reinforcement learning