MP neuron diagram for f(x1, x2, x3, x4) = (!x1 AND !x2) AND (x3 AND x4) with inhibitory inputs x1, x2, excitatory inputs x3, x4, threshold theta, output y in {0,1}, plus explanatory text Based on the above data, answer the given subquestions. Figure from the original question paper

(D) Option figure from the original question paper

(A) Option figure from the original question paper

(B) Option figure from the original question paper

(C) Option figure from the original question paper

September 2024 term · Deep Learning · BSCS3004

Deep Learning Quiz 1: 27 October 2024 (September 2024 term)

The IIT Madras BS Deep Learning (Deep Learning) Quiz 1 paper sat on 27 Oct 2024, in the September 2024 term: 12 questions for 40 marks in 120 minutes. Every question is below with its answer. Take it as a timed mock test to be marked, or read it through first.

Questions: 12
Marks: 40
Duration: 120 min
Numerical: 5
MCQ: 6
MSQ: 1

Take as mock test Practise with answers

Updated 25 Sept 2026

Official paper: IIT M DEGREE AN EXAM QDB2 27 Oct 2024 · No negative marking.

Question 1

+3 marksNumerical answer

Based on the above data, answer the given subquestions.

Show answer

Correct answer: 2

Question 1 on its own page

Question 2

+2 marksOne correct option

Based on the above data, answer the given subquestions.

Show answer

Correct answer

Question 2 on its own page

Question 3

+2 marksOne correct option

Based on the above data, answer the given subquestions.

Show answer

Correct answer

Question 3 on its own page

Question 4

+3 marksOne correct option

Based on the above data, answer the given subquestions.

Show answer

Correct answer

Question 4 on its own page

Question 5

+4 marksOne correct option

Consider a Perceptron model which outputs( $\hat{y}$ ) 1 if $\mathbf{w^T x} \ge 0$ else it outputs 0. This model has been trained on the given dataset. The learned weights are $\mathbf{w_1} = \mathbf{3}$ and $\mathbf{w_2} = -\mathbf{1}$ , corresponding to the two input features.

Based on this information, what is the possible range for the bias term $\mathbf{w_0}$ such that the Perceptron correctly classifies all the data points?

$x_1$	$x_2$	$y$
0	0	0
0	1	0
1	0	1
1	1	1

Show answer

Correct answer

Question 5 on its own page

Question 6

+3 marksOne correct option

A team has a data set that contains 20000 samples for training a feed-forward neural network. Suppose they used a mini-batch gradient descent algorithm with a batch size of 500 to update the weights. How many times do the weights get updated after training the network for 10 epochs?

A
400
B
20000
C
100000
D
40

Show answer

Correct answer

A
400

Question 6 on its own page

Question 7

+2 marksOne correct option

What is the key difference between Nesterov Accelerated Gradient (NAG) and traditional momentum-based gradient descent?

A
NAG uses an exponentially weighted moving average of gradients.
B
NAG calculates the gradient at the current point rather than a look-ahead point.
C
NAG calculates the gradient at a look-ahead point.
D
NAG only works for convex optimization problems.

Show answer

Correct answer

C
NAG calculates the gradient at a look-ahead point.

Question 7 on its own page

Question 8

+3 marksOne or more correct options

Which of the following strategies are commonly used for adjusting the learning rate during training?

Select all that apply.

A
Step decay
B
Momentum based gradient descent optimization
C
Stochastic gradient descent optimization
D
Exponential decay

Show answer

Correct answers

A
Step decay
D
Exponential decay

Question 8 on its own page

Question 9

+5 marksNumerical answer

Consider a sigmoid neuron that takes in an input vector $\mathbf{x} = \begin{bmatrix} 1 \\ 0.5 \\ 1 \end{bmatrix}$ . The weight vector $\mathbf{w}$ is initialized to $\begin{bmatrix} 0.5 \\ 1 \\ 0.5 \end{bmatrix}$ and $\mathbf{b} = 1$ . The output from the sigmoid neuron is

$\hat{\mathrm{y}} = \frac{1}{1 + exp^{-(\mathbf{w^T x} + \mathbf{b})}}$

Suppose we use the following loss function

$\mathcal{L} = \frac{1}{2}(y - \hat{y})^2$

Update the weight vector once by running the vanilla Gradient Descent algorithm with $\eta = 2$ . Assume the true label $\mathbf{y} = \mathbf{0}$ . What is the new loss value (that is, the loss computed after updating the weight vector)?