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

Training a model from scratch on a new dataset

Answer

Transferring data between different storage systems

Answer

Using multiple GPUs to speed up training

Question 2

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

Accepted Answer

ResNet

Answer

LSTM

Answer

Word2Vec

Answer

ARIMA

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

Adjusting hyperparameters such as batch size and epoch count

Answer

Pruning unused neurons from the network

Answer

Selecting which features to include in the input

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

The final classification layer

Answer

The batch normalization layers

Answer

All layers including the output layer

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

A preprocessing step that normalizes input data

Answer

A method for selecting the most informative training examples

Answer

A technique for compressing model weights

Question 6

What is the primary advantage of transfer learning over training a model from scratch?

Accepted Answer

It reduces the amount of labeled data and compute required to achieve good performance

Answer

It always produces higher accuracy regardless of dataset size

Answer

It eliminates the need for labeled data entirely

Answer

It prevents overfitting in all scenarios

Question 7

Which strategy is generally recommended when fine-tuning a pre-trained model?

Accepted Answer

Use a lower learning rate than was used in original pre-training to avoid destroying learned features

Answer

Use a high learning rate to quickly adapt all layers

Answer

Randomly reinitialize all weights before fine-tuning

Answer

Only train the input embedding layers