Enroll. No.
_________
MARWADI UNIVERSITY
Faculty of Engineering and Technology
Information and Communication Technology
B.Tech. SEM: V MU FINAL EXAM December 2023
______________________________________________________________________
Subject: - Machine Learning (01CT1519) Date:- 04/12/2023
Total Marks:-50 Time: - 120 minutes
Question 1: Do as directed: [CO1,CO2] [10]
1. If you are a machine learning engineer working in the industry, how will you select supervised [02]
algorithm. Also give the reason of selection.
2. Supervised algorithm has training and testing phases. Is there any algorithm which does not have [02]
training phase? Justify your answer.
3. What do you mean by outliers ? Which of the unsupervised learning algorithm works well in the [02]
present of outliers ? How ?
4. Demonstrate non linear boundary of EX-OR gate. Show EX-OR and AND gate realization with [02]
help of neural network and weights.
5. Draw the box plot for the given statistics of the dataset. Comment on the nature of the features, [02]
based on the box-plot obtained.
Question 2: Solve the following [Attempt any two]: [18]
1. For the given set of points identify clusters using single linkage and draw the Dendrogram. How [09]
many clusters are good to consider in this case ?
2. For the sunburn data set below construct decision tree. [09]
Name Hair Height Weight Location Class
Avni Blonde Average Light No Yes
Ramesh Blonde Tall Average Yes No
Shailja Brown Short Average Yes No
Khushali Blonde Short Average No Yes
Amarpreetsingh Red Average Heavy No Yes
John Brown Tall Heavy No No
Yasin Brown Average Heavy No No
Namrata Blonde Short Light Yes No
3. Show by taking the dummy example with (not less than) 2 features, the requirement of the [09]
normalization of data, in predicting the output value. Explain how the model may behave
adversely, in case the normalization is not involved in the pre-processing step. [CO1, CO3]
Question 3: Do as directed [08]
1. For the given confusion matrix, calculate various performance parameters and comment on it. [02]
[CO4]
1042 12
True label
35 43
Predicted Label
2. Calculate various performance metrics for the linear regression. [CO4] [02]
Sr. No. Xi Yi Y (hat)
(size of Actual Estimated
house) (in Lac) (in Lac)
1 80 40 42
2 90 46 45
3 120 72 70
4 105 64 67
5 72 35 34
6 145 112 115
3. Explain the process of using “Random Forest” as [02]
a. Classification Problem
b. Regression Problem
How would you evaluate the designed ML model for both the cases? [CO1, CO4]
4. You are designing a deep learning system to detect driver fatigue in cars. It is crucial that your [02]
model detects fatigue, to prevent any accidents. Design the system that achieves the given
problem statement. Also state, which of the following is the most appropriate evaluation metric :
Accuracy, Precision, Recall, Loss Value. Explain your choice [CO4]
Question 4: Do as directed [Attempt any 3]: [09]
1. After visually inspecting the dataset for classifying the historical places, you realize that the [03]
training set only contains pictures taken during the day, whereas the validation set only has
pictures taken at night. Explain what is the issue and how you would correct it. [CO5]
2. Suppose you have to rotate an image. Image rotation is nothing but multiplication of image by a [03]
specific matrix to get a new transformed image.
consider one point in the image to rotate with co-ordinates as (1, 0) to a co-ordinate of (0, 1),
identify the kernel matrix for the given task? [CO5]
3. For any given CNN model, the accuracy vs epochs and loss vs epochs graphs are as given [03]
below.
Interpret the graph obtained after training the model. Is the trained model, optimized? If not,
what possible changes can be done in model architecture to improve the accuracy and loss
graphs? Draw the improved accuracy and loss graphs. [CO5]
4. Consider the summary of the designed sequential model. Draw the architecture of the mentioned [03]
sequential model with all the necessary model configurations. [CO5]
Question 5: Do as directed [Attempt any 1]: [05]
1. Explain the training, validation and testing dataset and their importance in designing any ML [05]
model. Explain the significance of regularization w.r.t. bias and variance observed for any
designed ML model considering overfitting and underfitting concepts. [CO3]
2. Explain the process of feature selection and process to remove the redundant features from the [05]
dataset. Why the redundant features are required to be removed for training the model. Explain
giving examples. [CO3]
---Best of Luck---
