Scribd
Upload
16 views

Sample

Uploaded by

umarani-rs
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
16 views

Sample

Uploaded by

umarani-rs
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
You are on page 1/ 2

1. Solve the NOT gate using single layer perceptron. Assume there are 2 input.

(CO1)
2. Compare and contrast overfitting and underfitting with Error rate graph. (CO1)
3. State the principle of regularization in Deep FFNN. (CO1)
4. Define activation function. List out the various types of activation function. (CO1)
5. Define dropout and batch normalization. (CO1)
6. Differentiate between Shallow and Deep Networks. (CO2)
7. Consider the input image size as 224 X 224 X 3, Filter size as 5 X 5, no of filters as 64
stride as 1 and no of zero padding as 0 Compute the output volume of this convolutional
layer. (CO2)
8. Name any two hyper parameter used in CNN and define how it will helps in spatial
arrangement and size of the output volume. (CO2)
9. Draw the block diagram of LSTM and explain its components. (CO2)

10. a. Calculate the linear regression coefficients and plot the line of regression with
deviation for each of the following data. Also, compute the Mean Squared Error. (10
marks) (CO1)
X 1 2 3 4 5 6 7
Y 9 8 10 12 11 13 14
10. b. Suppose we have data on 100 cases of myocardial infarction and 150 healthy
individuals (mi = 1 if MI, 0 otherwise) matched to the MI group by age and sex. From their
medical records before the MI (if they had one), we classify the individuals as diabetic,
metabolic disorder, and normal blood glucose (bg = norm, metdis, diabetic). The table on the
next page shows the number of individuals in each group. (6 marks) (CO1)

a. Find the odds ratio of MI for diabetic individuals vs. normal individuals.
b. Find the odds ratio of Control for diabetic individuals vs. normal individuals.
c. Compute changes in Odds due to diabetic.
(or)

11. Given a neural network, find out the actual output vector [y1, y2] for the sample input
[0.05, 0.10]. Use the eights and bias terms as per the figure below. Form the Weight
Matrices, and bias vectors. Assume that the hidden nodes are designed with SGN(x)
activation function, where SGN(x) = 1, if x>0, SGN(x) = 0, otherwise. (16) (CO1)

12. Consider applying the convolutional Neural Network (CNN) below to a database of
images. Each image is of size 100×100 with 3 channels. That is, the input to the CNN
has the dimension of 100×100×3 and the output has 10 classes. All convolution filters
have stride 1 and all pooling functions have stride 2. Find the total number of parameters
in each layer? Also draw the architecture for the same. (16 marks) (CO2)

Output
Softmax - 10
FC1: Fully Connected
P2: 2×2 max pooling
C3: Conv 3×3-20, pad 2
P1: 2×2 max pooling
C2: Conv 5×5 – 10, pad 2
C1: Conv 3×3 – 10, pad 1
Input 100×100×3

(or)
13.b. Build a RNN-Language model for prediction of the occurrence of the next word, based
on probability distribution of ‘t’ past words. Show the architecture flow, for the sentence “The
Clouds are in the ____<w>_____”. Compute overall loss of this model. (8 marks) (CO2)

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy