USN 204CS001

Ph.D./M.S. (Engg.) by Research Degree Examination, December 2022

Machine Learning Techniques
Time: 3 hrs. Max. Marks:100
Note: Answer any FIVE full questions, choosing ONE full question from each module.
2. Any revealing of identification, appeal to evaluator and /or equations written eg, 42+8 = 50, will be treated as malpractice.

Module-1
1 a. Enlist and explain the various successful applications of Machine Learning. (06 Marks)
b. With a flow diagram, explain the final design of checkers learning program. (10 Marks)
c. Give the questions that form the issues in machine learning. (04 Marks)

OR
Important Note : 1. On completing your answers, compulsorily draw diagonal cross lines on the remaining blank pages.

2 a. With an example, explain in detail the working of FIND-S Algorithm. (10 Marks)
b. Write the Candidate-Elimination algorithm using version spaces. (06 Marks)
c. Enlist the characteristics that problems must have to apply Decision Tree. (04 Marks)

Module-2
3 a. Define the characteristics that problems should posses to apply Back propagations algorithm
used in ANN. (06 Marks)
b. Define and represent perceptrons. Why some of Boolean functions connot be represented by
single perceptron (06 Marks)
c. Write and explain Gradient descent algorithm differentiate standard and Stochastic gradient
descent. (08 Marks)

OR
4 a. Explain the common heuristics to attempt to alleviate. The problem of local minima.
(06 Marks)
b. Define Genetic algorithm and write a prototypical genetic algorithm. (09 Marks)
c. Explain with examples typical Genetic Algorithm operators for manipulating bit string
hypothesis. (05 Marks)

Module-3
5 a. Explain the features of Bayesian learning. Give and explain the significance of Baye’s
theorem. (08 Marks)
b. Design and explain a Brute – Force MAP learning algorithm to output maximum a posterior
hypothesis. (08 Marks)
c. Explain the mechanism of Bayes optimal classifier in answering the most probable
classification of new instances. (04 Marks)

OR
6 a. Write Naïve Bayes algorithm for learning and classifying text. (06 Marks)
b. Define the following :
i) True error ii) Training error iii) Sample error
iv) PAC-learnable v) Sample complexity. (10 Marks)
c. Explain K-TERM DNF and K-CNF concepts. (04 Marks)

1 of 2
 204CS001

Module-4
7 a. Mathematically express K-NEAREST NEIGHBOR algorithm for approximating a discrete
valued function f : Rn → V, Explain the refinement of distance weights. (08 Marks)
b. Explain locally weighted linear Regression with constraints. (08 Marks)
c. Enlist the genetic properties of case based reasoning systems that distinguish from
K-NEAREST NEIGHBOR. (04 Marks)

OR
8 a. Write and explain a prototypical sequential covering algorithm. (08 Marks)
b. Explain the various evaluation functions for evaluating. The performance used to guide
search in LEARN-ONE-RULE. (06 Marks)
c. Define :
i) Literal ii) Ground literal iii) Term iv) Horn clause v) Substitution vi) Antecedent.
(06 Marks)

Module-5
9 a. Explain the reasons for learning perfect domain theory. Write the explanation based learning
algorithm PROLOG – EBG. (08 Marks)
b. Explain the key properties of explanation based learning algorithm and its several
perspectives. (08 Marks)
c. Define the various methods for using prior knowledge to alter search performed. (04 Marks)

OR
10 a. Explain in detail working of FOCL algorithm. (10 Marks)
b. Bringout the differences between Reinforcement learning problems from other function
approximation task. Write Q-Learning algorithm. (10 Marks)

*****

2 of 2