Question 11
UnclassifiedTransfer learning typically involves which of the following approaches?
Correct answer: B
Explanation
Transfer learning means taking a model already trained on a large dataset and adapting it to a new task. Reusing a pre-trained model such as "ResNet, EfficientNet, ViT" and "fine-tuning it on a target task" works because the model has already learned useful features that can be updated for the new problem.
Why each option is right or wrong
A. Training a model from random initialization on every new dataset without using prior learned weights
B. Reusing a pre-trained model (e.g., ResNet, EfficientNet, ViT) and fine-tuning it on a target task
Transfer learning is the standard practice of starting from a model already trained on a source dataset and then adapting its learned weights to a new, related task; in deep learning this is commonly done by taking architectures such as ResNet, EfficientNet, or ViT and fine-tuning them rather than training from scratch. The key diagnostic feature is that the pre-trained parameters are retained and updated on the target data, often with the final classification head replaced or retrained, which is why this option matches the definition of transfer learning.
C. Using only the test set to train and evaluate the model
D. Removing all pretrained weights and discarding learned features before training