Scribd Logo
Upload

Welcome to Scribd!

  • Unit 4 - Week 3: Assignment 3

    Uploaded by

    Raushan Kashyap
    47 views3 pages

    Original Title

    W3A1

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Flag for inappropriate content
    Download as pdf or txt
    47 views3 pages

    Unit 4 - Week 3: Assignment 3

    Uploaded by

    Raushan Kashyap

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Flag for inappropriate content
    Download as pdf or txt
    of 3

    25/05/2018 Deep Learning For Visual Computing - - Unit 4 - Week 3

    [email protected] ▼

    Courses » Deep Learning For Visual Computing

    Announcements Course Ask a Question Progress Mentor

    Unit 4 - Week 3

    Course
    outline Assignment 3
    The due date for submitting this assignment has passed. Due on 2018-02-14, 23:59 IST.
    How to access
    the portal Submitted assignment

    Week 1 1) Which of the following models can be employed for unsupervised learning 1 point

    Week 2
    Both

    Week 3
    Deep Belief Networks
    Lecture 11 :
    Autoencoder for
    Representation
    Learning and Autoencoder
    MLP
    Initialization

    Lecture 12 : None
    MNIST
    handwritten
    No, the answer is incorrect.
    digits Score: 0
    classification Accepted Answers:
    using
    Both
    autoencoders

    Lecture 13 : 2) Which of the following statements are true for a autoencoder (neglect bias): 1 point
    Fashion MNIST
    classification (i) Number of input perceptrons are equal to number of output perceptrons
    using (ii) It needs supervised labels
    autoencoders
    (iii) It is a regression algorithm
    Lecture 14 :
    ALL-IDB (i), (ii) and (iii)
    Classification
    using
    (ii) and (iii)
    autoencoders

    Lecture 15 :
    Retinal Vessel
    (i) and (iii)
    Detection using
    autoencoders
    (i) and (ii)
    Week 3: Lecture
    Slides

    Quiz : No, the answer is incorrect.


    Assignment 3 Score: 0

    Week 3: Accepted Answers:


    Assignment (i) and (iii)
    Solutions
    3) What is the dimension of encoder weight matrix of an autoencoder (hidden units=400) 1 point
    Week 4 constructed to handle 10 dimensional input samples

    Week 5 rows = 10 and columns = 401

    https://onlinecourses.nptel.ac.in/noc18_ee08/unit?unit=7&assessment=75 1/3
    25/05/2018 Deep Learning For Visual Computing - - Unit 4 - Week 3

    rows = 11 and columns = 400


    Week 6
    rows = 400 and columns = 11
    Week 7 rows = 400 and columns = 10

    No, the answer is incorrect.


    Week 8
    Score: 0

    Week 9 Accepted Answers:


    rows = 400 and columns = 11
    Week 10
    4) What is the dimension of decoder weight matrix of an autoencoder (hidden units=400) 1 point
    constructed to handle 10 dimensional input samples
    Week 11
    rows = 11 and columns = 400
    Week 12
    rows = 400 and columns = 11
    rows = 10 and columns = 401
    DOWNLOAD
    VIDEOS rows = 400 and columns = 10

    No, the answer is incorrect.


    Score: 0
    Accepted Answers:
    rows = 10 and columns = 401

    5) Which of the following cases is best suited for training SAE and transferring those weights to 1 point
    initialize deep neural network for classification

    Ample unlabelled data and limited labelled data


    Big data and GPU
    Ample labelled data
    Any of the above

    No, the answer is incorrect.


    Score: 0
    Accepted Answers:
    Ample unlabelled data and limited labelled data

    6) Which of the following statement is true for a trained SAE (n=3) with E1, E2 and E3 being 1 point
    autoencoders

    E1 is hierarchically more representative


    Number of E3 weights should be greater that E1
    E3 is hierarchically more representative
    None

    No, the answer is incorrect.


    Score: 0
    Accepted Answers:
    E3 is hierarchically more representative

    7) What should En (nth block) represent for an ideal SAE trained on face dataset 1 point

    Edges
    Eigen face
    Curves
    None of the above

    No, the answer is incorrect.


    Score: 0
    Accepted Answers:
    Eigen face

    8) While exporting DNN as feature extractor for patch classification what should be the last 1 point
    block?

    Logistic regression
    Random forest

    https://onlinecourses.nptel.ac.in/noc18_ee08/unit?unit=7&assessment=75 2/3
    25/05/2018 Deep Learning For Visual Computing - - Unit 4 - Week 3

    Support vector machine


    Any of the above

    No, the answer is incorrect.


    Score: 0
    Accepted Answers:
    Any of the above

    9) What should E1 (1st encoder block) represent for an ideal SAE (stacked autoencoder) 1 point
    trained on MRI dataset

    Pathology
    Organs
    Edges
    Corners

    No, the answer is incorrect.


    Score: 0
    Accepted Answers:
    Edges

    10)Which loss function is used while training autoencoder 1 point

    Any classification loss


    Hinge loss
    Logistic regression
    Any regression loss

    No, the answer is incorrect.


    Score: 0
    Accepted Answers:
    Any regression loss

    Previous Page End

    © 2014 NPTEL - Privacy & Terms - Honor Code - FAQs -


    A project of In association with

    Funded by

    Powered by

    https://onlinecourses.nptel.ac.in/noc18_ee08/unit?unit=7&assessment=75 3/3

    You might also like