Quiz Space

Deep Learning for Computer Vision · Quiz 2 · 3 Aug 2025 · May 2025 term

Question 41: α22: _

Question 41

+0.5 marksNumerical answer

You are given the encoder hidden states h1,h2,h3∈R2h_1, h_2, h_3 \in \mathbb{R}^2 and the decoder hidden state s2∈R2s_2 \in \mathbb{R}^2 at time step t=2t = 2.

Use **dot-product attention**, where the attention score is computed as:

etj=st⊤hjandαtj=exp⁡(etj)∑k=13exp⁡(etk)e_{tj} = s_t^\top h_j \quad \text{and} \quad \alpha_{tj} = \frac{\exp(e_{tj})}{\sum_{k=1}^{3} \exp(e_{tk})}

The context vector is then computed as:

ct=∑j=13αtjhjc_t = \sum_{j=1}^{3} \alpha_{tj} h_j

Use the following values:

s2=[10],h1=[12],h2=[01],h3=[23]s_2 = \begin{bmatrix} 1 \\ 0 \end{bmatrix}, \quad h_1 = \begin{bmatrix} 1 \\ 2 \end{bmatrix}, \quad h_2 = \begin{bmatrix} 0 \\ 1 \end{bmatrix}, \quad h_3 = \begin{bmatrix} 2 \\ 3 \end{bmatrix}

Note: Use approximate values of exponential exp⁡(1)≈2.72\exp(1) \approx 2.72, exp⁡(2)≈7.39\exp(2) \approx 7.39, etc.

Answer the given subquestions:

α22: _________________

Show answer

Correct answer: 0.095 (accepted within ±0.005)

Question 41 of 52 in the IIT Madras BS Deep Learning for Computer Vision (Deep Learning for Computer Vision) Quiz 2 paper sat on 3 Aug 2025, in the May 2025 term (IIT M IMPROVEMENT AN EXAM QIB2 03 Aug 2025). It carries 0.5 marks.

More questions from this paper

  1. Q1Figure question
  2. Q2Which of the following architectures introduced the ROI pooling layer for extracting fixed-size feature maps from varia…
  3. Q3Which object detection method first integrated a Region Proposal Network (RPN) to generate region proposals within the …
  4. Q4Which architecture performs object detection as a single regression problem from image pixels to bounding boxes and cla…
  5. Q5Figure question
  6. Q6Consider the following statements P and Q regarding AlexNet and choose the correct option: (P) In AlexNet, a trainable …
  7. Q7Consider two bounding boxes in an image:\ Box A: top-left at (0, 0), bottom-right at (12, 12) - Box B: top-left at (4, …
  8. Q8Which one of the following statements is True?
  9. Q9Figure question
  10. Q10Match the following: | 1) One-to-many RNN architecture | i) Sentiment analysis of Netflix reviews | |---|---| | 2) Many…
  11. Q11Figure question
  12. Q12Which of the following techniques help mitigate the vanishing gradient problem in recurrent neural networks?\ Gradient …
  13. Q13Which of the following techniques help mitigate the exploding gradient problem in recurrent neural networks?\ Gradient …
  14. Q14Which one of the following statements is false? (Pick the most appropriate one.)
  15. Q15Which of the following is true regarding Hard Attention and Soft Attention?
  16. Q16Which of the following are true:
  17. Q17Consider a Faster R-CNN object detection framework, which is being trained with a mini-batch size of 2. The RPN produce…
  18. Q18Suppose you want to build an encoder-decoder recurrent model with attention. The context vector at time step t is given…
  19. Q19Suppose you want to build an encoder-decoder recurrent model with attention. The context vector at time step t is given…
  20. Q20Consider a Gated Recurrent Unit (GRU) cell at time step ’t’. For each of the entities in the given subquestions, enter …
  21. Q21Consider a Gated Recurrent Unit (GRU) cell at time step ’t’. For each of the entities in the given subquestions, enter …
  22. Q22Consider a Gated Recurrent Unit (GRU) cell at time step ’t’. For each of the entities in the given subquestions, enter …
  23. Q23Consider a Gated Recurrent Unit (GRU) cell at time step ’t’. For each of the entities in the given subquestions, enter …
  24. Q24Consider a Gated Recurrent Unit (GRU) cell at time step ’t’. For each of the entities in the given subquestions, enter …
  25. Q25Consider a Gated Recurrent Unit (GRU) cell at time step ’t’. For each of the entities in the given subquestions, enter …
  26. Q26Number of Parameters:_
  27. Q27Computational Cost:_
  28. Q28Number of parameters for depthwise seperable convolution:_
  29. Q29Computational Cost for depthwise seperable convolution:
  30. Q30Number of parameters for pointwise convolution:_
  31. Q31Computational cost for pointwise convolution:
  32. Q32Number of weights in Weight Matrix U1 is:
  33. Q33Number of weights in Weight Matrix V1 is:
  34. Q34Number of weights in Weight Matrix U2 is:
  35. Q35Number of weights in Weight Matrix V2 is:_
  36. Q36Number of weights in Weight Matrix W is:_
  37. Q37e21: _
  38. Q38e22: _
  39. Q39e23: _
  40. Q40α21: _
  41. Q42α23: _
  42. Q43c21: _
  43. Q44c22: _
  44. Q45e21:_
  45. Q46e22:_
  46. Q47e23:_
  47. Q48α21:_
  48. Q49α22:_
  49. Q50α23:_
  50. Q51c21:_
  51. Q52c22:_