Geeks of Coding

Join us on Telegram

Home Forums Assignment NPTEL_ Practical Machine learning with tensor flow Practical Machine Learning with Tensorflow – Assignment 1

Viewing 0 reply threads
  • Author
    • #403
      Abhishek TyagiAbhishek Tyagi

      1)- From the below graphs, select the one that satisfies the equation y=σ(0.1∗x) :


      2)- Perform z-score normalization and min-max scaling on the given array and select from the given options.

      arr = [ 100 , 50 , 400 , 300 , 100 ]

      Answer- None of the above

      3)- We have a neural network with an input layer of h0 nodes, hidden layers of h1, h2, h3…., h1 nodes respectively and an output layer of h(l+1) nodes. How many parameters does the network have?


      \sum \limits_{i=0}^l [(h_i*h_{i+1})+h_{i+1}]

      4)- Suppose we need to minimize the following loss function by tuning scalar value w using gradient descent: f(w) = 9 + 4w + w2 . Given w<sub>0</sub> =0, select the best value for learning rate α, such that gradient descent reaches the optimal value in just one step.

      Answer- 0.5

      5)- In linear regression with MAE loss, if we only had one target Y_train for all the observations X_train, then the optimal value that the prediction Y_pred should have is:

      Answer- Median of Y_train

      6)- A machine learning model gets an accuracy of 90% on a dataset with 90% positive class and 10% negative class. Can we conclude that the model is a good classifier of the data?

      Answer- No

      7)- Suppose you are given with the following training data for linear regression

      x = [3, 2, 4, 0]
      y = [4, 1, 3, 1]
      You are using a loss function
      J(w, b) = \frac{1}{2} \sum \limits_{i=1}^n [h(x_i) - y_i ] ^2
      What is the value of J(1,1)?

      Answer- 4

      8)- In the above question, if (w, b) = (1, 3), what is <i> h(4)</i> ?

      Answer- 7

      9)- If we halve the value of a given feature, what happens to the coefficients of other features estimated by minimizing squared loss function? (assuming no interaction between any two features)

      Answer- Stay the same

      10)- Which of the following model complexity vs. loss function plots is most likely from training data?


Viewing 0 reply threads
  • You must be logged in to reply to this topic.