DEEP LEARNING MCQ

1. Which neural network has only one hidden layer between the input and output?

1. Deep neural network

2. Feed-forward neural networks
3. Recurrent neural networks
4. Shallow neural network

Answer: Shallow neural network

2. Deep learning algorithms are more accurate than machine learning

algorithms in image classification

1. 0.37%
2. 0.41%
3. 33%
4. 0.4%

Answer: 0.41%

3. How many layers of Deep learning algorithms are constructed?

1. 3
2. 4
3. 2
4. 5

Answer: 4

4. RNNs stands for?

1. Receives neural networks

2. Recurrent neural networks
3. Recording neural networks
4. None pf above

Answer: Recurrent neural networks

5. Which of the following is/are Limitations of deep learning?

1. Obtain huge training datasets

2. Data labeling
3. None of the above
4. All of above

Answer: All of above

6. Which of the following is well suited for perceptual tasks?

 1. Recurrent neural networks
2. Convolutional neural networks
3. Reinforcement Learning
4. Feed-forward neural networks

Answer: Convolutional neural networks

7. CNN is mostly used when there is an?

1. unstructured data
2. structured data
3. None of the above
4. All of above

Answer: structured data

8. Which of the following functions can be used as an activation function in the output
layer if we wish to predict the probabilities of n classes (p1, p2..pk) such that the sum of
p over all n equals to1?

1. ReLu
2. Softmax
3. Tanh
4. Sigmoid

Answer: Softmax

9. In a neural network, knowing the weight and bias of each neuron is the most
important step. If you can somehow get the correct value of weight and bias for each
neuron, you can approximate any function. What would be the best way to
approach this?

1. Search every possible combination of weights and biases till you get the best value
2. Iteratively check that after assigning a value how far you are from the best values,
and slightly change the assigned values values to make them better
3. Assign random values and pray to God they are correct
4. None of these

Answer: Iteratively check that after assigning a value how far you are from the best
values, and slightly change the assigned values values to make them better

10. Which of the following is/are Common uses of RNNs?

1. Detect fraudulent credit-card transaction

2. Provide a caption for images
3. Businesses Help securities traders to generate analytic reports
4. All of the above

Answer: All of the above

11. In a simple MLP model with 8 neurons in the input layer, 5 neurons in the hidden
layer, and 1 neuron in the output layer. What is the size of the weight matrices
between the hidden output layer and the input hidden layer?

1. [5 x 1] , [8 X 5]
2. [1 X 5] , [5 X 8]
3. [8 X 5] , [ 1 X 5]
4. [8 X 5] , [5 X 1]

Answer: [5 x 1] , [8 X 5]

12. Which of the following techniques perform similar operations as a dropout in a

neural network?

1. Boosting
2. Bagging
3. Stacking
4. None of above

Answer: Bagging

13. Which of the following statement(s) correctly represents a real neuron?

1. A neuron has multiple inputs but a single output only

2. A neuron has a single input and a single output only
3. All of the above
4. A neuron has a single input but multiple outputs

Answer: All of the above

14. Which of the following would have a constant input in each epoch of training a
Deep Learning model?

1. Weight between input and hidden layer

2. Weight between hidden and output layer
3. Biases of all hidden layer neurons
4. Activation function of output layer

Answer: Weight between input and hidden layer

15. The number of nodes in the input layer is 10 and the hidden layer is 5.
The maximum number of connections from the input layer to the hidden
layer
are

1. more than 50
2. less than 50
3. it is an arbitrary value
4. 50

Answer: 50
16. Assume a simple MLP model with 3 neurons and inputs=1,2,3. The weights of
the input neurons are 4,5, and 6 respectively. Assume the activation function is a
linear constant value of 3. What will be the output?

1. 64
2. 128
3. 32
4. 96

Answer: 96

17. The input image has been converted into a matrix of size 28×28 and a kernel/filter
of size 7×7 with a stride of 1. What will be the size of the convoluted matrix?

1. 2
2. 21×21
3. 20×20
4. 25×25

Answer: 22×22

18. What consists of the Boltzmann machine?

1. asynchronous operation
2. fully connected network with both hidden and visible units
3. stochastic update
4. All of the above

Answer: All of the above

19. Which, if any, of the following propositions is true about fully-connected

neural networks(FCNN)?

1. A FCNN with only linear activations is a linear network.

2. In a FCNN, there are connections between neurons of a same layer
3. In a FCNN, the most common weight initialization scheme is the zero initialization,
because it leads to faster and more robust training.
4. None of above

Answer: A FCNN with only linear activations is a linear network

20. In which of the following applications can we use deep learning to solve
the problem?

1. Prediction of chemical reactions

2. Protein structure prediction
3. Detection of exotic particles
4. All of the above

Answer: All of the above

21. SVM stands for

1. Support Vector Mechanism

2. Super Visual Machine
3. Support Vector Machine
4. Support Vector Model

Answer: Support Vector Machine

22. Which statement is true?

1. Deep learning is an analogue framework for learning representations from data

2. Deep learning is a mathematical framework for learning representations from data
3. Deep learning is a biological framework for learning representations from brain
4. Deep learning is a digital framework for learning representations from data

Answer: Deep learning is a mathematical framework for learning representations from

data

23. Three technical forces are driving advances in machine learning

1. Pen and a piece of paper

2. Hardware, Datasets & benchmarks, and Algorithmic advances
3. Super computers only
4. All of the above

Answer: Hardware, Datasets & benchmarks, and Algorithmic advances

24. Which of the following statements is true when you use 1×1 convolutions in a CNN?

1. It can help in dimensionality reduction

2. It suffers less overfitting due to small kernel size
3. It can be used for feature pooling
4. All of the above

Answer: All of the above

25. In which neural net architecture, does weight sharing occur?

1. Recurrent Neural Network

2. Convolutional neural Network
3. Fully Connected Neural Network
4. Both 1 and 2

Answer: Both 1 and 2

26. Which of the following is a subset of machine learning?

1. Numpy
2. Deep Learning
 3. Scipy
4. All of the above

Answer: Deep Learning

27. The two key ideas of deep learning for computer vision

1. Support Vector Machines and loss function

2. Convolutional neural networks and backpropagation
3. Deep neural networks and kernel functions
4. None

Answer: Convolutional neural networks and backpropagation

28. What is loss function in deep learning?

1. To control the output of a neural network, you need to be able to measure how far
this output is from what you expected
2. To calculate loss in banks
3. These are the predicted values only
4. These are true targets of data

Answer: To control the output of a neural network, you need to be able to measure how
far this output is from what you expected

29. What is shallow learning in deep learning?

1. Machine learning tend to focus on learning 512 layers of representations of the data
2. Machine learning tend to focus on learning only one or two layers of representations
of the data
3. Machine learning tend to focus on learning 64 layers of representations of the data
4. Machine learning tend to focus on learning 10 layers of representations of the data

Answer: Machine learning tend to focus on learning only one or two layers of
representations of the data

30. What is representation in deep learning?

1. It gets closer to the expected output

2. It is a way to look at data to represent or encode
3. RGB and HSV are two different examples of representations
4. All of the above

Answer: All of the above

31. Computers are best at learning

(A) facts.
(B) concepts.
(C) procedures.
(D) principles.
Correct Answer: A
32. Supervised learning and unsupervised clustering both require at least one

(A) hidden attribute.

(B) output attribute.
(C) input attribute.
(D) categorical attribute.
Correct Answer: A

33. Supervised learning differs from unsupervised clustering in that supervised learning
requires

(A) at least one input attribute.

(B) input attributes to be categorical.
(C) at least one output attribute.
(D) ouput attributes to be categorical.
Correct Answer: B

34. Another name for an output attribute.

(A) predictive variable

(B) independent variable
(C) estimated variable
(D) dependent variable
Correct Answer: B

35. Classification problems are distinguished from estimation problems in that

(A) classification problems require the output attribute to be numeric.

(B) classification problems require the output attribute to be categorical.
(C) classification problems do not allow an output attribute.
(D) classification problems are designed to predict future outcome.
Correct Answer: C

36. Which statement is true about prediction problems?

(A)The output attribute must be categorical.

(B) The output attribute must be numeric.
(C) The resultant model is designed to determine future outcomes.
(D) The resultant model is designed to classify current behavior.
Correct Answer: D

37. Which statement about outliers is true?

(A) Outliers should be identified and removed from a dataset.

(B) Outliers should be part of the training dataset but should not be present in the test data.
(C) Outliers should be part of the test dataset but should not be present in the training data.
(D) The nature of the problem determines how outliers are used.
(E) More than one of a,b,c or d is true.
Correct Answer: D

38. Assume that we have a dataset containing information about 200 individuals. One hundred
of these individuals have purchased life insurance. A supervised data mining session has
discovered the following rule:
IF age < 30 & credit card insurance =
 yes THEN life insurance = yes
Rule Accuracy: 70%
Rule Coverage: 63%
How many individuals in the class life insurance no have credit card insurance and are less than
30 years old?
(A) 63
(B) 70
(C) 30
(D) 18
Correct Answer: D

39. Which statement is true about neural network and linear regression models?

(A) Both models require input attributes to be numeric.

(B) Both models require numeric attributes to range between 0 and 1.
(C) The output of both models is a categorical attribute value.
(D)Both techniques build models whose output is determined by a linear sum of weighted input
attribute values.
(E) More than one of a,b,c or d is true.
Correct Answer: A

40. Which of the following is a non-iterative method to generate adversarial

examples?

(i) Non-Saturating Cost Method

(ii) Input Optimization Method
(iii) Adversarial Training
(iv) Logit Pairing
(v) Fast Gradient Sign Method
(vi) Real-time Cryptographic Dropout Method

Solution: (v)

41. Batch normalization helps to prevent-

A. activation functions to become too high or low

B. the training speed to become too slow
C. Both A and B
D. None
Correct Answer: C

42. LSTM is a variation of-

A. Convolutional Neural Network

B. Recurrent Neural Network
C. Multi Layer Perceptron Network
D. None
Correct Answer: B

43. Which of the following is true about dropout?

A. Applied in the hidden layer nodes

B. Applied in the output layer nodes
C. Both A and B
D. None
Correct Answer: A

44. For an image classification task, which of the following deep learning algorithm is best
suited?
A. Recurrent Neural Network
B. Multi-Layer Perceptron
C. Convolution Neural Network
D. All of the above

Correct Answer: C

45. Which of the following is a deep learning library?

A. Tensorflow
B. Keras
C. PyTorch
D. All of the above

Correct Answer: D