Questions and published explanations below are available without starting a test. Some questions may not have a published solution yet.
0 / 37 published solutions
Question 152 NAT · 2.0 marks
In a semantic role labeling evaluation, a model has identified 100 correct semantic roles out of 120
total semantic roles it predicted. What is the precision of the model as a decimal (rounded to two
decimal places)?
A published solution is not available for this question yet.
What is the size of the feature map after applying two successive convolution operations with
given parameters? Image size = 64 x 64, Kernel size = 3 x 3, Padding = 2 and Stride = 3. (In
calculation, take floor(x) whenever x is non-integer. If the answer is FxF, write F in the blank).
A published solution is not available for this question yet.
Let an input to a convolutional layer in a CNN have size Df × Df × M where Df = 32 and M = 100, and
output feature map (after passing input through conv layer) has Df × Df × N size where N = 64. Let
the kernel in the conv layer be k × k where k = 5. Calculate the number of parameters for this
convolution layer. (Assume appropriate padding is applied for all convolutions so the input and
output sizes are equal. Ignore the bias term in the calculation)._________________
A published solution is not available for this question yet.
Identify the correct sequence of steps in a Canny edge detection pipeline. Steps are listed below:
1. Compute gradient magnitude and direction
2. Connect individual components
3. Smoothen the image
4. Threshold into strong, weak, or no edge
5. Gaussian Filter and Hysteresis
6. Non-maximum suppression
7. Apply derivative to get edges
6406533822701.
6→ 1→ 4→ 5 → 2
6406533822702.
3→ 1→ 6→ 4 → 2
6406533822703.
3→ 5→ 1→ 4 → 2
6406533822704.
6→ 1→ 5→ 7 → 2
A published solution is not available for this question yet.
Suppose we have a codebook of 2048 SIFT visual words. We extract 1000 SIFT descriptors (SIFT is a
128-dimensional feature) from a new image. What is the dimensionality of the BoW (Bag-of-Words)
descriptor?
6406533822726.
1000
6406533822727.
128
6406533822728.
128 × 2048
6406533822729.
2048
A published solution is not available for this question yet.
Consider a Convolutional Neural Network which processes an RGB image. It has 128 kernels each
of spatial dimension 5 × 5 in the first layer. They are convolved with a stride 1. This is followed by a
max-pooling layer with stride 2 and kernel size 5 × 5. What would be the receptive field size of a
single neuron in the pooling layer? (Recap: A receptive field is the size of the region in the input
image, which influences the activation of that specific neuron.)
6406533822740.
5 × 5
6406533822741.
7 × 7
6406533822742.
9 × 9
6406533822743.
3 × 3
A published solution is not available for this question yet.
Consider the following two statements. Which of the following statements are true?
(a) Convolution operator is both commutative and associative.
(b) Fourier transform of a convolved image FT (a * b), is not the product of the Fourier transform of
the constituent images FT(a) x FT(b)
6406533822744.
a and b
6406533822745.
not a but b
6406533822746.
a but not b
6406533822747.
Neither a nor b
A published solution is not available for this question yet.
A 3x3 kernel generates an output image of dimension 20x20 after convolution. The (approximate)
number of computations performed to obtain this output image is:
6406533822748.
3600
6406533822749.
180
6406533822750.
400
6406533822751.
9
A published solution is not available for this question yet.
Momentum in optimization can cause oscillations around minima when
encountering flat regions or saddle points due to the dominance of the momentum term over the
small gradient.
6406533822681.
Stochastic Gradient Descent (SGD) with its inherent noise can be beneficial in
escaping local minima and saddle points, provided that there is enough gradient information in
the neighborhood.
6406533822682.
Adagrad is an optimization algorithm introduced to overcome the diminishing
learning rate problem in techniques like RMSProp.
6406533822683.
ADAM is introduced to solve problems in RMSProp by combining RMSProp and
Adagrad techniques.
A published solution is not available for this question yet.
