Deep Learning Transfer Learning and Fine-Tuning

Question 1

What is transfer learning in deep learning?

Accepted Answer

Reusing knowledge from a model trained on one task to improve performance on a different task

Answer

Transfer learning leverages knowledge (weights and representations) acquired from a source task to improve learning efficiency and performance on a related target task.

Question 2

Which of the following is a pre-trained model commonly used for transfer learning in computer vision?

Accepted Answer

ResNet

Answer

ResNet (Residual Network) is a widely used pre-trained convolutional neural network originally trained on ImageNet, making it a standard backbone for computer vision transfer learning.

Question 3

What does 'fine-tuning' mean in the context of transfer learning?

Accepted Answer

Continuing to train some or all layers of a pre-trained model on a new dataset

Answer

Fine-tuning involves taking a pre-trained model and continuing its training on a new (often smaller) dataset so it adapts to the target task while retaining prior knowledge.

Question 4

Which layers are typically frozen during the initial stages of transfer learning?

Accepted Answer

The early convolutional layers that capture low-level features

Answer

Early layers learn general low-level features (edges, textures) that are broadly useful, so they are frozen to preserve this generic knowledge while later task-specific layers are updated.

Question 5

What is a 'feature extractor' in the context of transfer learning?

Accepted Answer

Using a pre-trained network's intermediate representations as input to a new model without updating the pre-trained weights

Answer

When using a pre-trained model as a feature extractor, its weights are frozen and its intermediate activations serve as fixed feature representations fed into a newly trained classifier.

Deep Learning Practice Test

Deep Learning Practice Test

Deep Learning Transfer Learning and Fine-Tuning