Q.1 What is the output of print(np.array([1,2,3]) +1)?
A. [11 21 31]
B. [1 1 2 3]
C. [1 2 3 1]
D. [2 3 4]
Ans : [2 3 4]
Q.2 Which of the below formula is used to update weights while performing gradient descent?
A. dw – learning_rate*w
B. w – learning_rate*dw
C. w /learning_rate*dw
D. w +learning_rate*dw
Ans : w – learning_rate*dw
Q.3 Cost is equal to average of sum of losses.
A. True
B. False
Ans : True
Q.4 In dot product the number of rows in first matrix must be equal to number of columns in second.
A. True
B. False
Ans : False
Q.5 What is the output of print(np.dot([1,2,3],[[1],[2],[3]])?
A. [[1 2 3] [2 4 6] [3 6 9]]
B. throws error
C. [14]
D. [[14]]
Ans : [14]
Q.6 What does it mean if derivatives of parameters with respect to cost is negative?
A. Current parameter value must be reduced
B. The cost function has reached its minimum
C. Current parameter value must be increased
D. None of the options
Ans : Current parameter value must be increased
Q.7 You are building a binary classifier for classifying output(y=1) vs. output(y=0). Which one of these activation functions would you recommend using for the output layer?
A. tanh
B. relu
C. sigmoid
D. leaky-relu
Ans : sigmoid
Q.8 What is the equation for linear output of a hidden_layer in shallow neural network, if X is of shape (num_features, num_samples) and W is of shape(num_neurons, num_input)?
A. Z = W.X+b
B. Z = transpose(W).X +b
C. Z = W.transpose(X) + b
D. Z = X.W + b
Ans : Z = W.X+b
Q.9 If a shallow neural network has five hidden neurons with three input features what would be the dimension of bias matrix of hidden layer?
A. (1,1)
B. (5,3)
C. (1,5)
D. (5,1)
Ans : (5,1)
Q.10 Hidden layer must use activation function with larger derivative.
A. True
B. False
Ans : True
Q.11 In shallow neural network, number of rows in weight matrix for hidden layer is equal to number of nodes (neurons) in hidden layer.
A. True
B. False
Ans : True
Q.12 A vector of size (n,1) is called a row vector.
A. True
B. False
Ans : False
Q.13 If a shallow neural network has five hidden neurons with three input features, what would be the dimension of weight matrix of hidden layer?
A. (5,5)
B. (3,3)
C. (5,3)
D. (3,5)
Ans : (5,3)
Q.14 If a shallow neural network has five hidden neurons with three input features what would be the dimension of bias matrix of hidden layer?
A. (5,3)
B. (5,1)
C. (1,1)
D. (1,5)
Ans : (5,1)
Q.15 For a single neuron network, if number of features is 5 then what would be the dimension of bias vector?
A. (5,5)
B. (1,1)
C. (1,5)
D. (5,1)
Ans : (1,1)
Q.16 How many hidden layers are present if layer_dims = [3,9,9,1]?
A. 1
B. 2
C. 3
D. 4
Ans : 1
Q.17 If layer_dims = [3,9,9,1], then the shape of weight vector for third layer is _____________.
A. (9,3)
B. (5,5)
C. (5,3)
D. (3,5)
Ans : (5,3)
Q.18 What is the output of print(np.array([1,2,3]) * np.array([1,2,3]) )?
A. [14 14 14]
B. [14]
C. [1 4 9]
D. [1 2 3 1 2 3]
Ans : [1 4 9]
Q.19 In dot product the number of rows in first matrix must be equal to number of columns in second.
A. True
B. False
Ans : False
Q.20 If layer_dims = [3,9,9,1], then the shape of weight vector for third layer is _____________.
A. (9,3)
B. (9,1)
C. (9,9)
D. (3,9)
Ans : (9,9)
Q.21 Hidden layer must use activation function with a larger derivative.
A. True
B. False
Ans : True
Q.22 Broadcasting in Python throws error when you try to add two vectors of shape(1,5) and (1,6).
A. True
B. False
Ans : True
Q.23 In case of DNN weight vector for each layer must always be initialized to zero before training the network.
A. True
B. False
Ans : False
Q.24 In a shallow neural network, number of rows in weight matrix for hidden layer is equal to number of nodes (neurons) in hidden layer.
A. True
B. False
Ans : True
