Convolutional Neural Networks (CNN)
1. Why is data augmentation used while training a CNN on the CIFAR-10 dataset?
2. What is the main advantage of using convolutional layers instead of fully connected layers for images?
3. Why is normalization applied to CIFAR-10 images before training the CNN?
4. What is the purpose of pooling layers in a CNN architecture?
5. Why is ReLU commonly used as an activation function in CNNs?
6. What role does Global Average Pooling play in the CNN model used in this experiment?
7. Why is the CIFAR-10 dataset considered challenging for CNN models?
8. What information does a confusion matrix provide in CNN evaluation?
9. Why are training and testing accuracy curves plotted after CNN training?
10. Which factor most strongly contributes to CNN's ability to generalize well in this experiment?