PSNA COLLEGE OF ENGINEERING & TECHNOLOGY
Dindigul – 624 622.
(An Autonomous Institution Approved by AICTE and Affiliated to Anna University, Chennai)
DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING
Academic Year: 2024-25 Regulation: 2022
Degree & Branch: M.E., CSE Date:
Subject Code & Name: CP1213 & Machine Learning Year & Sem.: I & II
Assignment Questions
Sl. No Answer All Questions. (2 * 16marks + 1 * 8mark = 40 marks)
Marks
CO
BL
Adaptive Learning Platform
An ed-tech company builds a learning platform that recommends personalized content
based on user behavior, such as quiz scores, time spent on topics, and navigation patterns.
The system uses the EM algorithm to cluster learners, and applies Reinforcement
1.
Learning (both model-based and TD learning) to optimize recommendations over time. 8 3 3
a) Illustrate how the EM algorithm can be used to group students based on their
interaction patterns. How does this help in delivering personalized content? (4)
b) Demonstrate how Temporal Difference (TD) Learning can adjust recommendations
based on immediate feedback like quiz scores. Provide a brief example. (4)
Smart Health Monitoring System
A tech company develops a Smart Health Monitoring System using wearable sensor data
(e.g., heart rate, sleep patterns) to predict patient health. The system uses Naïve Bayes for
classification, Maximum Likelihood/MAP for parameter estimation, a Bayesian Belief
2.
Network (BBN) to model symptom dependencies, and Hidden Markov Models (HMMs) 16 3 4
to track health states over time.
a) Explain how Naïve Bayes classifies health status using sensor inputs.(8)
b) Apply Maximum Likelihood or MAP to estimate the most likely health condition
from observed data. (8)
Image Recognition for E-commerce
An e-commerce company is using Deep Learning with a Convolutional Neural Network
(CNN) to automatically categorize product images (e.g., electronics, clothing). The system
3.
is trained on labeled data to improve recommendation accuracy. 16 3 5
a) Apply CNN to classify product images. How would you set up a CNN for this task?
(8)
b) Demonstrate how CNN can be trained with labeled data to improve accuracy.(8)