Scribd
Upload
26 views4 pages

CCD 7104 Machine Learning 2

Revision

Uploaded by

EG209/108489/21 ISAAC DUNCAN MWENDWA
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
26 views4 pages

CCD 7104 Machine Learning 2

Revision

Uploaded by

EG209/108489/21 ISAAC DUNCAN MWENDWA
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

MERU UNIVERSITY OF SCIENCE AND TECHNOLOGY

P.O. Box 972-60200 – Meru-Kenya.


Tel: +254(0) 799 529 958, +254(0) 799 529 959, +254 (0)712 524 293
Website: www.must.ac.ke Email: [email protected]

University Examinations 2020/2021

FIRST YEAR FIRST SEMESTER EXAMINATIONS FOR MASTER OF SCIENCE IN


INFORMATION TECHNOLOGY

CCD 7104: MACHINE LEARNING

DATE: JULY 2021 TIME: 3 HOURS

INSTRUCTIONS: Answer Question ONE and any other Two questions.

QUESTION ONE (20 MARKS)

Assume we have a set of data from patients who have visited Meru Level 5 hospital since the
year 2010. A set of features (e.g., temperature, height) have been also extracted for each
patient. Our goal is to decide whether a new visiting patient has any of diabetes, heart disease,
or Cancer (a patient can have one or more of these diseases).

a) Supposedly we have decided to use a neural network to solve this problem and we
have two choices:- either to train a separate neural network for each of the diseases or
to train a single neural network with one output neuron for each disease, but with a
shared hidden layer. You are consulted to provide your feedback on this dilemma.
Which method would you advise for adoption and why? (Provide a detailed
comparison in addition to your stance) (10 Marks)
b) Some patient features are expensive to collect (e.g., brain scans) whereas others are
not (e.g., temperature). Therefore, we have decided to first ask our classification
algorithm to predict whether a patient has a disease, and if the classifier is 80%
confident that the patient has a disease, then we will do additional examinations to
collect additional patient features In this case. The team you lead has the option of
using either of the following classifiers:- neural networks, decision tree, or naive

Meru University of Science & Technology is ISO 9001:2015 Certified


Foundation of Innovations Page 1
Bayes. You are required to provide a report comparing the three classifiers and finally
recommend one to fit the problem with justification (10 Marks)

Question Two (20 Marks)

a) Compare and contrast between each of the following pairs of machine learning models;
a. Generativevs discriminative (4 Marks)
b. Parametric vs Non-parametric (4 Marks)

b) Classification is a predictive modeling problem that involves assigning a class label to


each observation. Consequently, classification models generate a predicted class,
which comes in the form of a discrete category; and thus for most practical
applications, a discrete category prediction is required in order to make a decision.
However, this is highly influenced by class imbalances of data
a. Explain what you understand by the term class imbalance (2 Marks)
b. Describe some consequences of using unbalanced classes in classification
(4 Marks)
c. Using appropriate examples, outline any two mechanisms of resolving the
class imbalance problem (4 Marks)
d. Cite four different application areas in which class imbalance is practical
(2 Marks)
QUESTION THREE (20 MARKS)

a) Machine learning employs different strategies to enable a model to learn. Briefly


explain each of the following strategies;
a. Supervised Learning (2 Marks)
b. Unsupervised Learning (2 Marks)
c. Reinforcement Learning (1 Marks)
b) Machine learning is a branch of artificial intelligence that deals with the scientific
study of algorithms and statistical models that computer systems use to perform a
specific task without using explicit instructions, relying on patterns and inference
instead. Explain three other domains that form part of machine learning (6 Marks)
c) The goal of machine learning to improve the performance P, on some task T over
experience E. using these metrics (P, T, E). Using these metrics identify and justify
any three application areas within which machine learning can be employed(9 Marks)

Meru University of Science & Technology is ISO 9001:2015 Certified


Foundation of Innovations Page 2
QUESTION FOUR (20 MARKS)

a) Suppose that we want to solve the problem of finding out what a good car is by using
genetic algorithms. Suppose further that the solution to the problem can be
represented by a decision tree as follows:

i. Illustrate what would be the appropriate chromosome design for the given
problem (4 Marks)
ii. Describe the Genetic Algorithm parameters that need to be defined to solve
the problem (6 Marks)
iii. Solve the problem applying a single round of the prototypical Genetic
Algorithm (5 Marks)
iv. Explain your answer in (iii) above by providing the pseudo code of the
algorithm (5 Marks)

QUESTION FIVE (20 MARKS)

a) The following diagram provides the necessary steps for solving a machine learning
problem. Study it and answer the questions that follow;

Meru University of Science & Technology is ISO 9001:2015 Certified


Foundation of Innovations Page 3
a. Explain what you understand by the terms
i. Feature extraction and reduction (3 Marks)
ii. Dimensionality reduction (2 Marks)
iii. Cross validation (2 Marks)
iv. Hyper-parameter Optimization (2 marks)
b. Using an appropriate machine learning prediction problem, discuss the algorithms that
you would utilize in solving the problem from the onset to prediction (11 Marks)

Meru University of Science & Technology is ISO 9001:2015 Certified


Foundation of Innovations Page 4

You might also like