Quiz Space

Deep Learning · End Term · 3 Sept 2023 · May 2023 term · Set QPE1-S2

Question 6: The logistic sigmoid function is defined as follows, f(x)…

Question 6

+3 marksNumerical answer

The logistic sigmoid function is defined as follows,

f(x)=11+exp(−(wx+b))f(x) = \frac{1}{1 + exp(-(wx + b))}

The parameters are initialized to w=0.5w = 0.5 b=0.5b = 0.5. Suppose the loss in defined as

L=12(f(x)−y)2L = \frac{1}{2}(f(x) - y)^2

where yy is the true value. Compute the gradient of ww for the following sample x=1,y=1x = 1, y = 1.

Show answer

Correct answer: -0.055 (accepted within ±0.01)

Question 6 of 17 in the IIT Madras BS Deep Learning (Deep Learning) End Term paper sat on 3 Sept 2023, in the May 2023 term (IIT M DEGREE ET1 EXAM QPE1 S2 03 Sep). It carries 3 marks.

More questions from this paper

  1. Q1Figure question
  2. Q2Consider the statement “the attention mechanism in RNN based Encoder- Decoder architecture helps the decoder to underst…
  3. Q3Figure question
  4. Q4Figure question
  5. Q5Suppose that we have a deep Feed Forward Fully Connected Neural Network. The network is observed to have a high varianc…
  6. Q7Figure question
  7. Q8Figure question
  8. Q9Consider a simple RNN for a binary sequence classification problem.
  9. Q10How many neurons are there in the network. Every neuron in the network has bias associated with it?\ Note: A neuron is …
  10. Q11How many learnable parameters (including bias) does the network have? Assume dropout regularization is applied.
  11. Q12The statement that, in general, the activation function of neurons in the network are not necessarily be the same acros…
  12. Q13Consider a sentence inside the quote “I may be wrong, and you may be right, and by an effort, we may get nearer to the …
  13. Q14Suppose all words in the vocabulary are represented using one-hot-encoded vector of size |V|. Then compute the ordered …
  14. Q15The update rule for the ADAM (Adaptive Moments) optimization algorithm is given below, \begin{aligned} m_t &= \beta_1 m…
  15. Q16Suppose the input is x (one hot representation of context word) and the corresponding label is y (one hot representatio…
  16. Q17Based on the above data, answer the given subquestions.