Professional Documents
Culture Documents
Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
Disable Immersive Reader
Points: 25/50
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 1/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
You have an input volume that is 63x63x16, and convolve it with 32 filters
that are each 7x7, using a stride of 2 and no padding. What is the output
volume?
16x16x32
29x29x32
29x29x16
56x56x32
CNNs, unlike fully-connected neural networks, have neurons that are only
“sparsely connected”. What does it imply?
Each activation in layer (k+1) depends on a small number of activations from layer
k.
Each filter works on only the depth slices from the previous layer
Suppose your input is a 30 by 30 color (RGB) image, and you are not using a
convolutional network. If the first hidden layer has 10 neurons, each one fully
connected to the input, how many parameters does this hidden layer have
(including the bias parameters)?
900
2700
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 2/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
27000
27010
TRUE
FALSE
Can't say
Batch Normalization.
Hyperbolic Tangent.
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 3/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
Add an additional border of zeros around the output of a convolutional layer as a down-
sampling operation.
Add an additional border of zeros around the output of a convolutional layer such that it
retains the input size.
Add an additional border of zeros around the input such that input size can be
retained.
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 4/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
10
If all the weights are set to zero instead of random initializations in NN for a
classification task, what can be an expected behaviour?
The NN will train. However, all the neurons will end up recognizing the same thing.
The NN will not train.
None of these.
11
Until some point, increasing it reduces the variance of the model significantly with‐
out significant addition of bias to the model.
Controls the trade-off between the need for the model to fit the training set well and also
have a large number of model parameters.
12
[2, 4.33, 5]
[0, 0, 0]
13
Computationally efficient
No saturation of outputs in positive region
Can’t say
14
It normalizes (changes) all the input before sending it to the next layer.
None of these.
15
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 6/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
The constant-term.
16
OR
AND
NOR
NAND
17
Neural networks
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 7/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
18
Changing the input variable by 1 unit always affects the output by 1 unit too.
Since it is univariate, we need to estimate one coefficient for modelling the data.
19
TRUE
FALSE
can't say
20
Moving only the support vectors around affects the separating hyperplane as well.
Can be used for both classification and regression.
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 8/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
he parameter C in the cost function for SVM controls the trade-off between mis‐
classification and regularization.
Sensitive to noise.
21
Deep Neural Networks with Stochastic Depth bypasses the subset of layers
when training using
Linear function.
Identity function.
Self-Attention Module.
22
Ones
Zeros
Sparse matrix where non-zero elements are extracted from a normal distribution
with mean = 0, std-dev = 0.01.
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 9/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
23
NO
YES
It depends on gradient descent but not error surface.
Can't Say
24
Assertion (P): The path taken by Stochastic Gradient Descent (SGD) towards
the minima always has low variance and faster convergence as compared to
Batch gradient descent.
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 10/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
25
Let’s say, you are using activation function X in hidden layers of neural
network. At a particular neuron for any given input, you get the output as
“-0.0001”. Which of the following activation function(s) could X represent?
ReLU
TanH
Leaky ReLU with α = -0.01
Sigmoid
26
Faster convergence.
27
Which of the following are true with respect to parameter sharing in CNNs?
Reduces overfitting
Allows gradient descent to set many parameters to zero, making sparse connections be‐
tween neurons of the CNN layers
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 11/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
While training a CNN for image classification on a dataset of animals, instead of initializing
the network with random weights, use weights from a pre-trained CNN trained on
ImageNet. These shared weights (parameters) help in training convergence.
Allows one feature detector to be used in multiple local regions in the input image
28
29
In the context of deep CNNs, generally, we aim to achieve the Bayes error
instead of absolute zero error, because of
The available training input samples may not have full information about target
samples.
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 12/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
Once zero error is achieved, the system may enter into an uncertainty state.
30
Input: 57x57x3
Output volume: 28x28x3
2352
9747
0
None
31
TRUE
FALSE
Can't say
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 13/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
32
The deeper layers of a neural network are typically computing more complex fea‐
tures of the input than the earlier layers.
The earlier layers of a neural network are typically computing more complex features of
the input than the deeper layers.
Most number of parameters in the CNN are usually present in the fully connected
layers
Every neuron in a convolution layer looks for patterns in different regions of the input
33
Select all the options which helps to alleviate the vanishing gradient
problem:
34
Number of iterations
Bias vectors
Weight matrices
Learning rate
Number of neurons in each hidden layer
35
Stack of five 3x3 conv (stride 1) layers has same effective receptive field as
.............. layer ?
One 7x7
One 11x11
One 13X13
None
36
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 15/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
37
4. If the prediction does not match the output, change the weights
1234
3412
3142
3214
38
Why does VGG-Net use a stack of small 3x3 filters instead of a single, high-
dimensional kxk filter?
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 16/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
39
Which of the following are true for a pooling layer when used in a
Convolutional Neural Network:
If object is spatially translated, fails to help the CNN to detect its class.
40
Both Assertion, Reason are TRUE, but Reason is NOT proper explanation of
Assertion.
Both Assertion, Reason are TRUE, and Reason is proper explanation of Assertion.
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 17/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
41
42
Data Augmentation.
Weight Sharing.
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 18/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
43
As a fan of the movie franchise: Star Wars, you decide to build a system for
tracking the movements of Master Yoda by natural language commands, as
shown in the figure below:
Gradients become large, and to prevent divergence you have to slow down the learning
rate – hence convergence is slow
Hidden units become highly activated and convergence is faster as weights are high from
the beginning itself
As long as weights are randomly initialized, gradient descent is not affected by small or
large values of initialized weights
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 19/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
44
In Stochastic gradient descent, how many training samples are used before
updating the weights?
One.
None of the above.
45
46
47
In the context of deep CNN, for a given arbitrary low-level vision task, a
CNN-based model is proposed. The model has a fixed kernel size
throughout. The authors claim that “increasing the kernel size will result in
significant performance gain”. What can you say about this claim?
TRUE.
FALSE.
48
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 21/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
49
The training data size is not large enough. Collect a larger training data and retrain
it.
Tune the learning rate and add regularization term to the objective function.
Use the same training data but add a few more hidden layers.
50
1 and 2.
Only 1.
Only 2.
None of these.
This content is created by the owner of the form. The data you submit will be sent to the form owner. Microsoft is
not responsible for the privacy or security practices of its customers, including those of this form owner. Never give
out your password.
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 22/23
09/09/2022, 21:22 Deep Learning (CS 590) MID SEMESTER EXAM, (July - November 2021) 25th SEPTEMBER (9-11AM)
https://forms.office.com/Pages/ResponsePage.aspx?id=jacKheGUxkuc84wRtTBwHCTd1h3zvShFq0ZkcCDdWudUOVNJUlUzWFZESUtSQlRX… 23/23