Deep Learning Week 11 Nptel Assignment Answers
Are you looking for Deep Learning Week 11 Nptel Assignment Answers ? You’ve come to the right place! Access the most accurate answers at Progiez.
Table of Contents
NPTEL Deep Learning Week 11 Assignment 11 Answers (Jan-Apr 2025)
Course Link: Click Here
1) Which of the following can be a target output of a semantic segmentation problem with 4 classes?
a)
b)
c)
d)
2) Suppose you have a 1D signal x = [1,2,3,4,5] and a filter f = [1,2,3,4], and you perform stride 2 transpose convolution on the signal x by the filter f to get the signal y. What will be the signal y if we don’t perform cropping?
3) What are the different challenges one faces while creating a facial recognition system?
a) Different illumination condition
b) Different pose and orientation of face images
c) Limited dataset for training
d) All of the above
4) Fully Connected Convolutional Network (FCN) became one of the major successful network architectures. Can you identify the advantages of FCN which make it a successful architecture for semantic segmentation?
a) Larger Receptive Field
b) Mixing of global features
c) Lesser computation required
d) All of the above
5) In a Deep CNN architecture, the feature map before applying a max pool layer with (2×2) kernel is given below:
a)
b)
c)
d)
NPTEL Deep Learning Week 11 Assignment 11 Answers
6) What could be thought of as a disadvantage of Fully Convolutional Neural Networks (FCN) for semantic segmentation addressed by other researchers?
a) It has a fixed receptive field, so objects with a lesser size than the receptive field will be missed by the network.
b) Downsampling the image dimension over the depth makes the feature map sparse.
c) It requires a lot of computation.
d) None of the above.
7) What will be the dice coefficient of the following two one-hot encoded vectors? (IAI = number of 1 bits)
a) 0.83
b) 0.41
c) 0.67
d) 0.90
8) What will be the value of the dice coefficient between A and B?
a) 0.93
b) 0.77
c) 0.11
d) 0.89
9) In FaceNet, why is the L2 normalization layer used?
a) To constrain the embedding function in a d-dimensional hyper-sphere.
b) For regularization of the weight vector, i.e., L2 regularization.
c) For getting a sparse embedding function.
d) None of the above.
10) What is the use of Skip Connection in image denoising networks?
a) Helping the deconvolution layer recover an improved clean version of the image.
b) Backpropagating the gradient to bottom layers, which makes training easy.
c) To create a direct path between the convolution layer and the corresponding mirror deconvolution layer.
d) All of the above.
These are NPTEL Deep Learning Week 11 Assignment 11 Answers
More weeks of Deep Learning: Click Here
More Nptel Courses: https://progiez.com/nptel
NPTEL Deep Learning Week 11 Assignment 11 Answers