Scribd
Upload
57 views2 pages

Neural Network Question Answer

The document contains a series of question and answer pairs related to neural networks, covering topics such as activation functions, loss functions, and optimization techniques. Key concepts include the use of ReLU and Sigmoid functions, the differences between Batch and Stochastic Gradient Descent, and the importance of dimensionality reduction in feature maps. It also addresses common issues like the vanishing gradient problem and provides fixes for input shape discrepancies in different frameworks.

Uploaded by

Jasvinder Kaur
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
57 views2 pages

Neural Network Question Answer

The document contains a series of question and answer pairs related to neural networks, covering topics such as activation functions, loss functions, and optimization techniques. Key concepts include the use of ReLU and Sigmoid functions, the differences between Batch and Stochastic Gradient Descent, and the importance of dimensionality reduction in feature maps. It also addresses common issues like the vanishing gradient problem and provides fixes for input shape discrepancies in different frameworks.

Uploaded by

Jasvinder Kaur
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

NEURAL NETWORK QUESTION ANSWERS

1. Answer: c) ReLU 23. Answer: b) To reduce the spatial


dimensions of the feature map
2. Answer: b) Multi-class
classification 24. Answer: b) Sigmoid

3. Answer: c) Once per epoch 25. Answer: b) 0.5

4. Answer: b) Binary Cross-Entropy 26. Answer: a, c

5. Answer: a) 3x3 27. Answer: a, b, c

6. Answer: b, d 28. Answer: a, b

7. Answer: a, b 29. Answer: 200

8. Answer: a, b, c, d 30. Answer: 0.4621

9. Answer: 1.0 31. Answer: b) Add dropout layers

10. Answer: 0 32. Answer: c) Dropout

11. Answer: a) Sigmoid → ReLU 33. Answer: tensor(0)

12. Answer: Mini-batch Gradient 34. Issue:


Descent
input_shape=(64, 64, 3) is incorrect in
13. Answer: (1, 4, 14, 14) PyTorch because PyTorch follows
(channels, height, width) format, while
14. Fix: Replace Sigmoid() with
TensorFlow uses (height, width,
LeakyReLU(negative_slope=0.1).
channels).
15. Answer: Tanh’s zero-centered
Fix:
output aids gradient flow during
backpropagation. Change input_shape=(64, 64, 3) to
input_shape=(3, 64, 64).

35. Answer: Batch Gradient Descent


updates weights after processing
16.
the entire dataset, while Stochastic
17. Correct Order: Convolution → Gradient Descent updates weights
MaxPooling → Flattening → Fully after each sample.
Connected

18. Answer: MSE and MAE


36.
19. Answer: Converts multi-
37. Answer: It converts the 2D feature
dimensional feature maps into 1D
maps into a 1D vector for input into
vectors for dense layers.
fully connected layers.
20. Answer: b) Chain rule
38. Answer: It reduces the spatial
21. Answer: d) Momentum Gradient dimensions of the output feature
Descent map.

22. Answer: a) ReLU is computationally 39. Answer: c) Huber Loss


efficient

1
NEURAL NETWORK QUESTION ANSWERS

40.
41. Answer: It allows the model to shift
the activation function for better
fitting.

42. Answer: To compute gradients of


the loss function with respect to the
weights.

43. Answer: It balances the efficiency


of Batch Gradient Descent and the
noise reduction of Stochastic
Gradient Descent.

44. Answer: Tanh

45. Answer: It controls the step size


during weight updates.

46. Answer: -0.1

47. Answer: It suffers from the


vanishing gradient problem.

48. Answer: 5x5

49. Answer: To convert raw scores into


probabilities for multi-class
classification.

50. Answer: It avoids the vanishing


gradient problem and is
computationally efficient.

You might also like