CNN Transfer Learning with CNNs 5

Question 1

Which transfer learning strategy is best when the target dataset is small and similar to the source dataset?

Accepted Answer

Freeze the convolutional base, train only the classifier

Answer

When datasets are similar and target data is limited, freezing the base preserves useful features while avoiding overfitting in the new classifier.

Question 2

What is the role of Global Average Pooling (GAP) in modern transfer learning architectures?

Accepted Answer

Reduces each feature map to a single value, replacing flatten layers

Answer

GAP averages each feature map spatially, reducing parameters and overfitting compared to large flatten-then-dense layers.

Question 3

In few-shot learning via transfer learning, what enables a CNN to generalize from very few examples?

Accepted Answer

Rich feature representations learned from large source datasets

Answer

Pretrained features encode rich visual knowledge that allows meaningful comparison even with only a few target examples per class.

Question 4

When applying transfer learning to object detection (e.g., using Faster R-CNN), what part of the network is typically pretrained?

Accepted Answer

The backbone CNN used as a feature extractor

Answer

The backbone CNN (e.g., ResNet) is pretrained on ImageNet and used to extract feature maps, while detection-specific heads are trained from scratch.

Question 5

Which technique involves training a lightweight student model to mimic a larger pretrained teacher model?

Accepted Answer

Knowledge distillation

Answer

Knowledge distillation trains a compact student to replicate the teacher's soft outputs, transferring knowledge to a smaller deployable model.

CNN - Convolutional Neural Networks Practice Test