Recent Posts

Coursera: Machine Learning (Week 4) Quiz - Neural Networks: Representation| Andrew NG

▸ Neural Networks - Representation :


  1. Which of the following statements are true? Check all that apply.
    • Any logical function over binary-valued (0 or 1) inputs x1 and x2 can be (approximately) represented using some neural network.

    • Suppose you have a multi-class classification problem with three classes, trained with a 3 layer network. Let be the activation of the first output unit, and similarly and . Then for any input x, it must be the case that .

    • A two layer (one input layer, one output layer; no hidden layer) neural network can represent the XOR function.

    • The activation values of the hidden units in a neural network, with the sigmoid activation function applied at every layer, are always in the range (0, 1).








  1. Consider the following neural network which takes two binary-valued inputs
    and outputs . Which of the following logical functions does it (approximately) compute?
    enter image description here
    • AND
      This network outputs approximately 1 only when both inputs are 1.

    • NAND (meaning “NOT AND”)

    • OR

    • XOR (exclusive OR)



  1. Consider the following neural network which takes two binary-valued inputs
    and outputs . Which of the following logical functions does it (approximately) compute?
    enter image description here
    • AND

    • NAND (meaning “NOT AND”)

    • OR
      This network outputs approximately 1 when atleast one input is 1.

    • XOR (exclusive OR)








  1. Consider the neural network given below. Which of the following equations correctly computes the activation ? Note: is the sigmoid activation
    function.
    enter image description here
    • Thiscorrectly uses the first row of and includes the “+1” term of .











  1. You have the following neural network:
    enter image description here
    You’d like to compute the activations of the hidden layer . One way to do
    so is the following Octave code:
    enter image description here
    You want to have a vectorized implementation of this (i.e., one that does not use for loops). Which of the following implementations correctly compute ? Check all
    that apply.
    • z = Theta1 * x; a2 = sigmoid (z);
      This version computes correctly in two steps , first the multiplication and then the sigmoid activation.

    • a2 = sigmoid (x * Theta1);

    • a2 = sigmoid (Theta2 * x);

    • z = sigmoid(x); a2 = sigmoid (Theta1 * z);



Check-out our free tutorials on IOT (Internet of Things):


  1. You are using the neural network pictured below and have learned the parameters (used to compute ) and (used to compute as a function of ). Suppose you swap the parameters for the first hidden layer between its two units so and also swap the output layer so . How will this change the value of the output ?
    enter image description here
    • It will stay the same.
      Swapping swaps the hidden layers output . But the swap of cancels out the change, so the output will remain unchanged.

    • It will increase.

    • It will decrease

    • Insufficient information to tell: it may increase or decrease.



Click here to see solutions for all Machine Learning Coursera Assignments.
&
Click here to see more codes for Raspberry Pi 3 and similar Family.
&
Click here to see more codes for NodeMCU ESP8266 and similar Family.
&
Click here to see more codes for Arduino Mega (ATMega 2560) and similar Family.

Feel free to ask doubts in the comment section. I will try my best to answer it.
If you find this helpful by any mean like, comment and share the post.
This is the simplest way to encourage me to keep doing such work.


Thanks & Regards,
- APDaga DumpBox

10 comments:

  1. Why I can't represent XOR function without hidden layers? If I have a case like question 2 but with the weights: -10, 20, -20 I would get:

    x1 | x2 | xor
    0 | 0 | 0
    0 | 1 | 1
    1 | 0 | 1
    1 | 1 | 0

    wouldn't I?

    ReplyDelete
    Replies
    1. NO.
      If you consider a NN as given in question 2, with weights as -10, 20, -20 with NN equation -10 + (20*x_1) - (20*x_2) > 0. where threshold for activation is 0.
      you will get output as follows:

      x1 | x2 | Output | (>0 or not)
      0 | 0 | -10 | 0
      0 | 1 | -30 | 0
      1 | 0 | 10 | 1
      1 | 1 | -10 | 1

      Which doesn't represent XOR at all.

      Delete
  2. please explain 2 one some clearly'i did not undertand what is the target of output.

    ReplyDelete
    Replies
    1. I can't understand what exactly you want to ask.

      Delete
  3. in 2nd question they ask Which of the following logical functions does it (approximately) compute?
    out put answer what shoud come to satisfy the truth table. In 2 nd one first bit you answered This network outputs approximately 1 only when both inputs are 1.In bit 2 also same.out put howmuch should come to satisfy the truth table.(that means -30,20,10)

    ReplyDelete
    Replies
    1. Even Though your question is not very clear to me, I am trying to answer it as per my understanding.
      1. In Q2(1), Ans is AND gate as Output is 1 when "both" the input are 1.
      2. In Q2(2), Ans is OR gate as Output is 1 when "at least one" input is 1.
      NOTE: Here we have considered activation threshold = 0.

      3. If in Q2, we consider weights as -30, 20, 10 then the ans is AND gate (considering activation threshold = -1)
      -30 | 20 | 10
      1 | x1 | x2 Output
      1 | 0 | 0 = -30 = 0
      1 | 0 | 1 = -20 = 0
      1 | 1 | 0 = -10 = 0
      1 | 1 | 1 = 0 = 1

      NOTE: If we consider activation threshold=0 here, then it will be ambiguous for x1=1, x2=1 case. It won't represent AND gate or any other logic gate.

      Delete
  4. Why question 1 option 2 is incorrect?

    ReplyDelete
    Replies
    1. The outputs of a neural network are not probabilities, so their sum need not be 1

      Delete
    2. True. But it represents classes. So exactly one class must be true for a training data. So sum of all the values must be 1. isn't it?

      Delete
  5. I didn't get the question numbers 4 and 5 can you please explain in detail?
    I mean how did writing the vectorized implementation here work in place of for loop? and similarly in question 5 how do I calculate whether output changes or remains same?

    ReplyDelete