Total No. of Questions : 8] SEAT No.

7
23
PC1844 [6353]-166 [Total No. of Pages :3

ic-
tat
T.E. (Mechanical)

4s
ARTIFICIAL INTELLIGENCE & MACHINE LEARNING

7:2
02 91
(2019 Pattern) (Semester- II) (302049)

9:3
0
40
Time : 2½ Hours] 5/1 13 [Max. Marks : 70
0
Instructions to the candidates:
1/2
.23 GP

1) Answer Q.1 or 2, Q.3 or 4, Q.5 or 6, Q.7 or 8.

2) Neat diagrams must be drawn wherever necessary.
E
72

7
3) Figures to the right indicate full marks.
C

23
ic-
4) Assume Suitable data wherever necessary.
16

tat
5) Use of electronic pocket calculator is allowed.
8.2

4s
.24

7:2
91
Q1) a) Explain the concept of a random forest classifier. Discuss how it works
49

9:3
and why it is considered a powerful ensemble learning technique. [8]
30
40

b) Discuss the limitations and challenges of the K-Means Clustering

01
02

algorithm. How does it work? Provide examples of scenarios where

1/2

K-Means may not perform well. [9]

GP
5/1

OR
CE
72

Q2) a) Explain the Principle & Working of Logistic Regression with neat Sketch

7
23
& State its Advantages & Disadvantages. [8]
.23

b) Given following dataset, Find out it is likely that Blue Jeans from Brand ic-
16

tat
X would be on Sale Using Rayes Classifier. [9]
8.2

4s

Colour Cloth Type Brand On Sale?

.24

7:2
91

Blue Jeans Y Yes

49

9:3

Blue Shirt X No
30
40

Blue Jeans Y Yes

01
02

Black Jeans Y Yes

1/2
GP

Black Jeans X No
5/1

Black Shirt X Yes

CE
72

Black Shirt Y Yes

.23

Blue Jeans Y No
16

Black Jeans X Yes

8.2

Blue Shirt X Yes

.24
49

P.T.O.
Q3) a) Explain the steps involved in the development of the ML (Classification

7
23
or Regression) model. [10]

ic-
b) Explain different performance evaluators used for interpretation/

tat
assessment of classification models. Explain 2×2 confusion matrix and

4s
explain its terminology. Explain Cohen’s Kappa, F Score, ROC Curve.[8]

7:2
02 91
9:3
i) Accuracy

0
40
ii) 5/1 13
Precision
0
1/2
.23 GP

iii) Recall

iv) F1 Score
E
72

7
C

23
OR

ic-
16

tat
Q4) a) From the below confusion matrix, determine accuracy, recall,precision,F
8.2

4s
1-score, True Positive Rate (TPR), False Positive Rate (FPR), True
.24

7:2
Negative Rate (TNR), and False Negative Rate (FNR). Interpret the
91
49

9:3
Result: [10]
30
40

Actual Values
01
02
1/2

1 0
GP
5/1

Predicted 1 540 150

CE
72

7
23
.23

Values 0 110 200

ic-
16

tat
8.2

4s
.24

7:2

b) What is meant by “Hyper parameter tuning” and how it is used to make

91

a machine learning algorithm work better? Explain tuning of Hyper

49

9:3

parameters for any one specific algorithm. [8]

30
40
01
02
1/2
GP

Q5) a) What is Convolutional Neural Network (CNN)?Explain activation

5/1

functions in CNN? [10]

CE
72

b) Explain working of deep learning with an example and list out its
.23

advantages and disadvantages. [8]

16
8.2

OR
.24
49

[6353]-166 2
Q6) a) A neuron with 4 inputs has the weights 0.3, 0.2, 01, 0.5 and bias 0. The

7
23
activation function is linear, say the function f(x) = 2x. If the inputs are

ic-
0.8, 0.3, 0.2, 0.1 compute output and compare it with a neuron with

tat
same input parameters but with tanh and sigmoid activation function and

4s
interpret the result. [10]

7:2
b) Write a short note on [8]

02 91
9:3
i) Markov Decision Process

0
40
ii) 5/1 13
Artificial Neural Network
0
1/2
.23 GP

Q7) a) Explain any one mechanical engineering application where image-based

E
72

7
classification can be adopted. [8]
C

23
ic-
b) Explain human and machine interaction. [9]
16

tat
8.2

OR

4s
.24

7:2
Q8) a) How AI & ML used for Material Inspection with example? [8]
91
49

9:3
b) Explain in detail how machine learning is used for fault detection in the
30
40

manufacturing industry. [9]

01
02
1/2
GP
5/1


CE
72

7
23
.23

ic-
16

tat
8.2

4s
.24

7:2
91
49

9:3
30
40
01
02
1/2
GP
5/1
CE
72
.23
16
8.2
.24
49

[6353]-166 3
Total No. of Questions : 8] SEAT No. :

8
23
P735 [Total No. of Pages : 3

ic-
[5870]-1026

tat
T.E. (Mechanical)

5s
1:5
ARTIFICIAL INTELLIGENCE & MACHINE LEARNING

02 91
8:3
(2019 Pattern) (Semester - II) (302049)

0
20
9/0 13
0
Time : 2½ Hours] [Max. Marks : 70
6/2
.23 GP

Instructions to the candidates:

1) Neat diagrams must be drawn wherever necessary.
E
82

8
2) Figures to the right side indicate full marks.
C

23
ic-
3) Use of calculator is allowed.
16

tat
4) Assume suitable data wherever necessary.
8.2

5s
.24

1:5
91
49

Q1) a) Explain following terms in decision tree : [6]

8:3
30

i) entropy
20
01
02

ii) information gain

6/2
GP

iii) Gini index

9/0

b) What is the difference betwen KNN and K means? Also state advantages
CE
82

8
and limitations of KNN and K means? [5]

23
.23

ic-
c) How does the Bayes algorithm differ from decision trees? [6]
16

tat
OR
8.2

5s

Q2) a) What is Supeort Vector Machine? How does the SVM work? [6]
.24

1:5
91
49

b) Define pruning. What are various types of pruning? Explain any one
8:3

type of pruning. [5]

30
20

c) Differentiate between logistic regression and linear regression. [6]

01
02
6/2
GP
9/0

Q3) a) What are different hyperparameter tuning algorithms? Elaborate using

CE
82

an example. [8]
.23

b) Why data pre-procesing is required? Explain the techniques in pre-

16

processing. [6]
8.2

c) State advantages and disadvantages of random forest. [4]

.24
49

P.T.O.
 OR

8
23
Q4) a) Explain the difference between training data and Testing data in a Dataset?

ic-
How it is useful in a Machine Learning Model? [8]

tat
b) Explain the following terms : [6]

5s
1:5
i) Over fitted model

02 91
8:3
ii) Underfitted model

0
20
iii) 9/0 13
Good model
0
c) Define following terms [4]
6/2
.23 GP

i) ACCURACY
E
82

8
ii) PRECISION
C

23
ic-
iii) RECALL
16

tat
iv) F1SCORE
8.2

5s
.24

1:5
91
49

Q5) a) What do you understand from on policy and off policy algorithm in
8:3
30

reinforcement learning? Explain SARSA algorithm for Reinforcement

20

learning. [8]
01
02

b) Explain with neat diagrame quivalence of biological neuron and artificial

6/2
GP

neuron? [6]
9/0

c) Compute the output of the following neuron if the activation function is

CE
82

8
sigmoid. Assume bias to be 2.1. [4]

23
.23

ic-
16

tat
8.2

5s
.24

1:5
91
49

8:3
30
20
01
02

OR
6/2

Q6) a) What are different activation functions? Explain any one in details. [8]
GP
9/0

b) Explain [6]
CE
82

i) Positive Learning
.23

ii) Negative Learning with respect to Reinforcement learning

16

c) What are applications of Reinforcement learning in Mechanical

8.2

engineering? [4]
.24
49

[5870]-1026 2
Q7) a) Write short note on use of AIML in material inspection. [8]

8
23
b) What are the advantages of using fault detection in Automobile cars.[5]

ic-
c) What are the different applications of AIML in health care. [4]

tat
5s
OR

1:5
Q8) a) Explain in detail different applications of AIML. [8]

02 91
8:3
b) Write short note on use of AIML in traffic control. [5]

0
20
c) 9/0 13
What are the different types of sensors used in Human machine
interactions? [4]
0
6/2
.23 GP


E
82

8
C

23
ic-
16

tat
8.2

5s
.24

1:5
91
49

8:3
30
20
01
02
6/2
GP
9/0
CE
82

8
23
.23

ic-
16

tat
8.2

5s
.24

1:5
91
49

8:3
30
20
01
02
6/2
GP
9/0
CE
82
.23
16
8.2
.24
49

[5870]-1026 3
Total No. of Questions : 8] SEAT No. :

8
23
P7660 [Total No. of Pages : 3

ic-
[6180]-184

tat
T.E. (Mechanical)

3s
ARTIFICIAL INTELLIGENCE & MACHINE LEARNING

7:3
02 91
(2019 Pattern) (Semester - II) (302049)

9:4
0
30
Time : 2½ Hours] [Max. Marks : 70
2/1 13
Instructions to the candidates:
0
1) Answer Q.1 or Q.2, Q.3 or Q.4, Q.5 or Q.6 and Q.7 or Q.8.
2/2
.23 GP

2) Neat diagrams must be drawn wherever necessary.

3) Assume suitable data if necessary.
E

4) Use of Non-Programmable Scientific Calculator is allowed.

81

8
C

23
ic-
16

tat
Q1) a) Use ID3 algorithm to construct a decision tree for the data in the following
8.2

3s
table: [6]
.24

7:3
Instance No. X1 X2 Class label
91
49

9:4
1 T T 1
30
30

2 T T 1
01
02

3 T F 0
2/2
GP

4 F F 1
2/1

5 F T 0
CE
81

8
23
6 F T 0
.23

b) How does the K Nearest Neighbour (KNN) algorithm works? ic-

[7]
16

tat
8.2

3s

c) Explain bagging and boosting. [4]

.24

7:3

OR
91
49

9:4

Q2) a) Use Naïve Bayes algorithm to determine whether a Red SUV Domestic
30
30

car is a stolen car or not using the following data: [6]

01
02

Example No. Colour Type. Origin Whether stolen

2/2
GP

1 Red Sports Domestic Yes

2/1

2 Red Sports Domestic No

CE
81

3 Red Sports Domestic Yes

.23

4 Yellow Sports Domestic No

16

5 Yellow Sports Imported Yes

8.2

6 Yellow SUV Improted No

.24

P.T.O.
49

[6180]-184 1
 7 Yellow SUV Imported Yes

8
23
8 Yellow SUV Domestic No

ic-
9 Red SUV Imported No

tat
10 Red Sports Imported Yes

3s
7:3
b) Explain working of Support Vector Machines? What are hyper parameters

02 91
in SVM? [7]

9:4
0
30
c) How Naïve Bayes is used for classification. [4]
2/1 13
0
2/2
.23 GP

Q3) a) What is hyper parameter tuning? Explain any two hyper parameters in the
Random Forest algorithm. [9]
E
81

8
b) What is a confusion matrix? Write any 2 × 2 confusion matrix showing,
C

23
True Positive, True Negative, False Positive, False negative. [9]

ic-
16

tat
Define
8.2

3s
i) Accuracy
.24

7:3
ii) Precision
91
49

9:4
iii) Recall
30
30

OR
01
02

Q4) a) Why is data preprocessing required? Explain techniques of preprocessing

2/2
GP

data. [9]
2/1
CE

b) Explain following performance evaluators used for interpretation/

81

8
23
assessment of classification model: [9]
.23

ic-
16

i) Cohen’s Kappa Coefficient

tat
8.2

3s

ii) F Score
.24

7:3
91

iii) ROC Curve

49

9:4
30
30
01
02

Q5) a) Define Markov property. Explain why Markov property is applicable in

2/2

solving Reinforecement learning. [9]

GP
2/1

b) The transfer function of neurons on one layer of a neural network is

CE
81

assumed to be of sigmoid form. Evaluate the output of neurons

corresponding to input x = 0.62. How is the nature of sigmoid function?
.23

(Justify the answer with plot) [9]

16
8.2
.24

OR
49

[6180]-184 2
Q6) a) What are the characteristics of deep learning? What is the difference

8
23
between deep learning and machine learning? [9]

ic-
b) Explain the working og biological neurons? Explain with a neat diagram

tat
equivalence of biological neuron and artificial neuron. [9]

3s
7:3
02 91
9:4
Q7) a) Explain any one mechanical engineering application where image-based

0
30
classification can be adopted.
2/1 13 [8]
b) What is predictive maintenance? Explain different steps in predictive
0
2/2
.23 GP

maintenance. [9]
E

OR
81

8
C

23
Q8) a) Explain human and machine interaction? [8]

ic-
16

tat
b) Write a short note on use of AIML in traffic control. [9]
8.2

3s
.24

7:3
91
49

9:4

30
30
01
02
2/2
GP
2/1
CE
81

8
23
.23

ic-
16

tat
8.2

3s
.24

7:3
91
49

9:4
30
30
01
02
2/2
GP
2/1
CE
81
.23
16
8.2
.24
49

[6180]-184 3
Total No. of Questions : 8] SEAT No. :

8
23
PB-3901 [Total No. of Pages : 3

ic-
[6262]-166

tat
1s
T.E. (Mechanical)

9:4
ARTIFICIAL INTELLIGENCE & MACHINE LEARNING

02 91
9:2
(2019 Pattern) (Semester - II) (302049)

0
40
5/0 13
Time : 2½ Hours] [Max. Marks : 70
0
5/2
Instructions to the candidates:
.23 GP

1) Answer Q1 or Q2, Q3 or Q4, Q5 or Q6, Q7 or Q8.

2) Neat Diagrams must be drawn wherever necessary.
E
81

8
3) Assume suitable data, if necessary.
C

23
4) Use of Non-Programmable Scientific Calculator is allowed.

ic-
16

tat
Q1) a) Is Naive Bayes supervised or unsupervised algorithm? Why? [2]
8.2

1s
b) Differentiate between clustering and classification. [6]
.24

9:4
91
c) Explain how Support Vector Machine works ? Explain with neat sketch
49

9:2
Hard Margin and Soft Margin. [9]
30
40

OR
01
02

Q2) a) Define following terms of Decision tree. [2]

5/2
GP

i) Leaf node
5/0

ii) Pruning
CE
81

8
b) How does K-means work? [6]

23
.23

ic-
c) Use Naive Bayes algorithm to determine whether a red domestic SUV
16

tat
car is a stolen car or not using the following data : [9]
8.2

1s

Example no. Colour Type Origin Whether stolen

.24

9:4

1 red sports domestic yes

91
49

9:2

2 red sports domestic no

30
40

3 red sports domestic yes

01
02

4 yellow sports domestic no

5/2
GP

5 yellow sports imported yes

5/0

6 yellow SUV imported no

CE
81

7 yellow SUV imported yes

.23

8 yellow SUV domestic no

16

9 red SUV imported no

8.2

10 red sports imported yes

.24
49

P.T.O.
Q3) a) What are four typical problems to be solved using machine learning

8
23
approach? [6]

ic-
b) Enlist and explain steps involved in development of classification model.

tat
[6]

1s
9:4
c) Explain use of Confusion matrix in Machine Learning Model with suitable

02 91
example. [6]

9:2
0
40
OR
5/0 13
Q4) a) What is hyper parameter tuning? Explain any three hyper parameters
0
5/2
.23 GP

tuned in SVM? [6]

b) What is training data, labeled data and unlabeled data? What are key
E
81

8
steps involved in developing training data? [6]
C

23
ic-
c) Explain with neat sketch K-fold cross-validation mode. [6]
16

tat
8.2

1s
.24

9:4
Q5) a) Explain the concept of Reinforcement learning with an example. Also
91
49

9:2
define key terms used in Reinforcement learning. [8]
30
40

b) Explain Q-learning algorithm with flow diagram. [6]

01
02

c) The transfer function of neuron on one layer of a neural network is

5/2
GP

assumed to be of sigmoid form. Evaluate the output of neuron

5/0

corresponding to input x = 0.62. How is the nature of sigmoid function?

CE
81

(Justify the answer with plot). [4]

8
23
.23

OR
ic-
16

tat
Q6) a) Explain Convolution Neural Network (CNN) using neat flow diagram.
8.2

1s

Explain padding and striding in CNN. [8]

.24

9:4
91

b) Explain SARSA algorithm for reinforcement learning. [6]

49

9:2
30

c) A neuron with 4 inputs has the weights 1,2,3,4 and bias 0. The activation
40

function is linear, say the function f(x) = 2x. If the inputs are 4, 8,5,6
01
02

compute the output. Draw a diagram representing the neuron. [4]

5/2
GP
5/0
CE
81

Q7) a) Explain human and machine interaction? Explain with any example. [5]
.23

b) What is predictive maintenance? Explain different steps in predictive

16

maintenance. [6]
8.2

c) Explain with suitable example how fault detection is done. [6]

.24
49

[6262]-166 2
 OR

8
23
Q8) a) Explain different steps in Dynamic system reduction. [5]

ic-
tat
b) Explain any one mechanical engineering application where image-based

1s
classification can be adopted. [6]

9:4
c) Explain the steps involved in material inspection? How machine learning

02 91
9:2
can be implemented in material inspection. [6]

0
40
5/0 13 
0
5/2
.23 GP
E
81

8
C

23
ic-
16

tat
8.2

1s
.24

9:4
91
49

9:2
30
40
01
02
5/2
GP
5/0
CE
81

8
23
.23

ic-
16

tat
8.2

1s
.24

9:4
91
49

9:2
30
40
01
02
5/2
GP
5/0
CE
81
.23
16
8.2
.24
49

[6262]-166 3
Total No. of Questions : 8] SEAT No. :

8
23
PA-1534 [Total No. of Pages : 3

ic-
[5926]-154

tat
T.E. (Mechanical Engineering)

7s
4:1
ARTIFICIAL INTELLIGENCE & MACHINE LEARNING

02 91
9:3
(2019 Pattern) (Semester - II) (302049)

0
30
4/0 13
Time : 2½ Hours] [Max. Marks : 70
0
1/2
.23 GP

Instructions to the candidates:

1) Answer Q1 or Q2, Q3 or Q4, Q5 or Q6, Q7 or Q8.
E
81

8
2) Neat diagram must be drawn wherever necessary.
C

23
3) Figures to the right side indicate full marks.

ic-
16

4) Use of calculator is allowed.

tat
8.2

7s
5) Assume suitable data wherever necessary.
.24

4:1
91
49

9:3
30
30

Q1) a) Explain following terms: [6]

01
02

i) Accuracy
1/2
GP
4/0

ii) Precision
CE
81

8
iii) Recall

23
.23

iv) F-1 Score ic-

16

tat
8.2

7s

b) Explain the procedure to find out the optimum value of K in K-means

.24

4:1

clustering? [5]
91
49

9:3
30

c) What is over-fitting in decision tree? Explain the techniques to avoid

30

over-fitting. [6]
01
02
1/2

OR
GP
4/0

Q2) a) Explain the steps in KNN algorithm. [6]

CE
81

b) What is SVM? How does it work? [5]

.23
16

c) Explain the evaluation parameters for regression model. [6]

8.2
.24
49

P.T.O.
Q3) a) Explain the steps involved in development of ML model. [7]

8
23
b) Quality Engineer wants to solve a two-class classification problem for

ic-
predicting whether a product is defective. The actual number of products

tat
containing no defect are 950 (Truly predicted positives = 900), the actual

7s
number defective products are 150 (Truly predicted negatives = 130).

4:1
So, calculate accuracy, precision, recall and f1 score. [4]

02 91
9:3
c) Explain hyperparameter tuining in decision tree. Why is it required? [7]

0
30
4/0 13
0
OR
1/2
.23 GP

Q4) a) What are the different cross validation techniques? Explain K-fold cross
E
81

validation with neat sketch. [7]

8
C

23
ic-
b) A sugar factory produces 3 sizes of sugar from three different nets.
16

Daily 1000 tons of sugar produced from net-1, 3000 tons produced

tat
8.2

from net-2 and 2000 tons produced from net-3. The last year season

7s
experience shows that 1.5% of the total sugar produced from net 1 is
.24

4:1
waste sugar. The corresponding fractions of waste sugars for the
91
49

9:3
remaining nets are 2.5% and 2% respectively. A certain amount of sugar
30

is taken as a sample at random and is found to be waste sugar. Find out

30

the probability that it is produced from : [4]

01
02
1/2
GP

i) Net 1
4/0

ii) Net 2
CE
81

8
23
.23

iii) Net 3
ic-
16

tat
c) What are the different classification algorithms? Explain logistic regression
8.2

7s

with neat sketch. [7]

.24

4:1
91
49

9:3
30

Q5) a) Explain the concept of Reinforcement learning with suitable example.

30

Define following terms in Reinforcement learning: [8]

01
02
1/2
GP

i) Agent
4/0
CE

ii) State
81
.23

iii) Environment
16

iv) Reward
8.2
.24

[5926]-154
49

2
 b) Define Markov property. Explain why Markov property is most applicable

8
23
in solving Reinforcement learning problems. [6]

ic-
c) The transfer function of neuron on one layer of a neural network is

tat
assumed to be of sigmoid form. Evaluate the output of neuron

7s
corresponding to input x = 0.62. How is the nature of sigmoid function?

4:1
02 91
(Justify the answer with plot). [4]

9:3
0
30
OR
4/0 13
0
Q6) a) Explain Convolution Neural Network (CNN) using neat flow diagram.
1/2
.23 GP

Explain padding and striding in CNN. [8]

E
81

8
C

23
b) Explain Q-learning algorithm with flow diagram. [6]

ic-
16

tat
c) A neuron with 4 inputs has the weights 1, 2, 3, 4 and bias 0. The activation
8.2

7s
function is linear, say the function f(x) = 2x. If the inputs are 4, 8, 5, 6
.24

4:1
compute the output. Draw a diagram representing the neuron. [4]
91
49

9:3
30
30
01
02

Q7) a) How deep learning can be used for image classification? [6]
1/2
GP
4/0

b) Explain human-machine interaction with suitable examples. [5]

CE
81

8
23
c) Explain in detail various applications of AI in mechanical engineering.[6]
.23

ic-
16

tat
OR
8.2

7s
.24

4:1

Q8) a) How ML can be used in predictive maintenance? [6]

91
49

9:3
30

b) Explain use of AI in fault detection. [5]

30
01
02

c) Explain applications of AI in healthcare sector. [6]

1/2
GP
4/0
CE
81


.23
16
8.2
.24

[5926]-154
49

3
Total No. of Questions : 8] SEAT No. :

8
23
P345 [Total No. of Pages : 3

ic-
[6003]-426

tat
T.E.(Mechanical)

2s
ARTIFICIAL INTELLIGENCE & MACHINE LEARNING

1:5
(2019 Pattern) (Semester - II) (302049)

02 91
0:3
Time:2½ Hours] [Max. Marks : 70

0
31
1/0 13
Instructions to the candidates:
1) Neat diagram must be drawn wherever necessary.
0
2) Figures to right side indicate full marks.
6/2
.23 GP

3) Use of calculator is allowed.

4) Assume suitable data wherever necessary.
E
82

8
C

23
ic-
Q1) a) Explain following terms: [6]
16

tat
i) Accuracy.
8.2

2s
ii) Precision.
.24

1:5
91
iii) Recall.
49

0:3
30

iv) F-1 Score.

31

b) Explain the procedure to find out the optimum value of K in K-means

01
02

clustering? [5]
6/2
GP

c) Explain the following terms: [6]

1/0

i) Entropy.
CE
82

8
23
ii) Information gain.
.23

OR ic-
16

tat
8.2

Q2) a) Explain the steps in KNN algorithm. [6]

2s
.24

b) What is SVM? How does it work? [5]

1:5
91
49

c) Explain the evaluation parameters for regression model [6]

0:3
30
31
01

Q3) a) Explain the steps involved in development of ML model. [7]

02
6/2

b) Quality Engineer wants to solve a two-class classification problem for

GP
1/0

predicting whether a product is defective. The actual number of products

CE

containing no defect are 950 (Truly predicted positives = 900), the actual
82

number defective products are 150 (Truly predicted negatives = 130).

.23

So, calculate accuracy, precision, recall and fl score. [4]

16

c) Explain hyperparameter tuning parameters in decision tree. [7]

8.2

OR
.24

P.T.O.
49

[6003]-426 1
Q4) a) What are the different cross validation techniques? Explain K-fold cross

8
23
validation with neat sketch. [7]

ic-
b) A sugar factory produces 3 sizes of sugar from three different nets. Daily

tat
1000 tons of sugar produced from net-1, 3000 tons produced from net-

2s
2 and 2000 tons produced from net-3. The last year season experience

1:5
shows that 1.5% of the total sugar produced from net 1 is waste sugar.

02 91
0:3
the corresponding fractions of waste sugars for the remaining nets are

0
31
2.5% and 2% respectively. A certain amount of sugar is taken as a sample
1/0 13
at random and is found to be waste sugar. Find out the probability that it
0
is produced from. [4]
6/2
.23 GP

i) Net 1
E
82

8
ii) Net 2
C

23
ic-
iii) Net 3
16

tat
c) What are the different classification algorithms? Explain logistic regression
8.2

2s
with neat sketch. [7]
.24

1:5
91
49

Q5) a) Explain the concept of Reinforcement learning with suitable

0:3
30

example.Define following terms in Reinforcement learning. [8]

31
01

i) Agent
02
6/2
GP

ii) State
1/0

iii) Environment
CE
82

8
23
iv) Reward
.23

ic-
16

b) Define Markov property. Explain why Markov property is most applicable

tat
8.2

in solving Reinforcement learning problems. [6]

2s
.24

1:5

c) The transfer function of neuron on one layer of a neural network is

91
49

assumed to be of sigmoid from. Evaluate the output of neuron

0:3
30

corresponding to input x = 0.62.How is the nature of sigmoid function?

31

(Justify the answer with plot). [4]

01
02
6/2

OR
GP
1/0
CE
82
.23
16
8.2
.24
49

[6003]-426 2
Q6) a) Explain Convolution Neural Network (CNN) using neat flow diagram.

8
23
Explain padding and striding in CNN. [8]

ic-
b) Explain Q-learning algorithm with flow diagram. [6]

tat
2s
c) A neuron with 4 inputs has the weights 1,2,3,4 and bias 0. The activation

1:5
function is linear, say the function f(x) = 2x. If the inputs are 4,8,5,6

02 91
compute the output. Draw a diagram representing the neuron. [4]

0:3
0
31
1/0 13
Q7) a) How deep learning can be used for Tuning of control algorithm? [6]
0
6/2
.23 GP

b) Explain AI based fault detection. [5]

E

c) Explain in detail various applications of AI in mechanical engineering.[6]

82

8
C

23
OR

ic-
16

tat
Q8) a) How AIML can be used in Dynamic system reduction? [6]
8.2

2s
b) Explain HMI with suitable examples. [5]
.24

1:5
91
49

c) Explain applications of AI in process optimization. [6]

0:3
30
31
01
02
6/2
GP


1/0
CE
82

8
23
.23

ic-
16

tat
8.2

2s
.24

1:5
91
49

0:3
30
31
01
02
6/2
GP
1/0
CE
82
.23
16
8.2
.24
49

[6003]-426 3