WorksheetsNLP_4
Total questions: 10
Worksheet time: 5mins
Name
Class
Date
1.
What is the primary purpose of an activation function in a neural network?
a)
A) To initialize the weights
b)
B) To calculate the loss
c)
C) To introduce non-linearity
d)
D) To perform gradient descent
2.
In gradient descent, what does the learning rate control?
a)
A) The size of the weight updates
b)
B) The number of layers in the network
c)
C) The number of training examples
d)
D) The activation function used
3.
Which loss function is commonly used for regression problems in neural networks?
a)
A) Cross-Entropy Loss
b)
B) Mean Squared Error (MSE)
c)
C) Categorical Cross-Entropy Loss
d)
D) Sigmoid Loss
4.
What is the primary role of the Chain Rule in neural network training?
a)
A) Initializing weights
b)
B) Calculating gradients
c)
C) Determining the learning rate
d)
D) Choosing the activation function
5.
What does the softmax activation function do in multi-class classification?
a)
A) Introduces non-linearity to smooth the Loss function
b)
B) Converts output of last hidden layer into probabilities
c)
C) Calculates the mean squared error
d)
D) Determines the learning rate
6.
What does the term "Black Box" refer to in the context of neural networks?
a)
A) A network with no hidden layers is called a black box
b)
B) The output layer of a network
c)
C) A network with no activation functions
d)
D) NN are not easily interpretable nor explainable
7.
In a neural network with multiple layers, what is the purpose of the Chain Rule when computing gradients?
a)
It simplifies the loss function
b)
B) It ensures convergence to a global minimum
c)
C) It calculates Gradients through backpropagation
d)
D) It prevents overfitting by adding regularization terms
8.
Given a binary classification problem where the true label is 1 and the predicted probability is 0.7, what is the Cross-Entropy Loss?
a)
A) log(0.3)
b)
B) -log(0.3)
c)
C) log(0.7)
d)
D) -log(0.7)
9.
In a multi-class classification problem with three classes (Class A, Class B, and Class C), if the true label is Class C and the predicted probabilities are [0.2, 0.7, 0.1], what is the Categorical Cross-Entropy Loss?
a)
A) log(0.1)
b)
B) -log(0.9)
c)
C) log(0.2)
d)
D) -log(0.1)
10.
In a neural network with a single weight w, if the loss function is L= yhat2+1 and yhat=4w ,. What's the Gradient of the Loss w.r.t w , at w=0.1
a)
3.2
b)
6.4
c)
4.8
d)
0.8
100 %
