AI-900 Practice Question: Describe fundamental principles of machine learning on Azure
What is 'transfer learning' and how is it different from training from scratch?
⚠ Common exam trap
Watch out — candidates often confuse transfer learning with simply reusing a model without any retraining, or think it only applies when original data is missing, rather than understanding it as a resource-efficient fine-tuning strategy.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Transfer learning fine-tunes a pre-trained model on a new task — requiring far less data and compute than training from scratch
Transfer learning starts with a model already trained on a large dataset (e.g., ImageNet) and fine-tunes it on a smaller, task-specific dataset. This approach requires significantly less data and computational resources compared to training from scratch, where all model weights are randomly initialized and learned from the ground up. It is especially effective when the new task is similar to the original training task, allowing the pre-trained features to be reused.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Transfer learning and training from scratch produce identical results
Why it's wrong here
Training from scratch initializes model weights randomly and forces the model to discover low-level features solely from the target dataset. Transfer learning instead initializes from weights that already encode general patterns from a large source dataset, so the fine-tuned model converges faster and usually achieves higher accuracy on small datasets. Thus the two approaches produce different, not identical, results.
- ✓
Transfer learning fine-tunes a pre-trained model on a new task — requiring far less data and compute than training from scratch
Why this is correct
Transfer learning takes a model already trained on a broad, large-scale dataset and fine-tunes it on a smaller, task-specific dataset, often with a low learning rate. The early layers retain generic features, while later layers adapt to the new task, so the model needs only a fraction of the data and compute that training from scratch would require. This is why pre-trained models can be adapted to custom scenarios with relatively few labeled examples.
- ✗
Transfer learning copies model weights between Azure subscriptions
Why it's wrong here
Moving a serialized model from one Azure subscription to another is an asset-management task handled by Azure Machine Learning models and registries, not an ML training method. Transfer learning operates at the mathematical level of neural network weights, reusing and fine-tuning representations learned on a source task for a new task. It has nothing to do with cloud subscription boundaries.
- ✗
Transfer learning is used only when the original training data is unavailable
Why it's wrong here
Transfer learning is not contingent on whether the original training data is missing; it can be beneficial even when a large target dataset exists. By starting from pre-trained weights, fine-tuning reduces computational cost and often yields better generalization than training from scratch. Its purpose is efficiency and performance, not data unavailability.
Go deeper
Related to this question
About these practice questions
This AI-900 question is part of Courseiva's 985-question bank — original exam-style content with full explanations and wrong-answer analysis, never real exam questions or exam dumps. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This AI-900 practice question is part of Courseiva's free Microsoft certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the AI-900 exam.