AIF-C01 Question 44
Select 3Your company needs to implement a machine learning model to classify images into different categories. The team is considering whether to use an open-source pre-trained model or to train a custom model. Which of the following factors should influence the decision?
- A
The availability of a pre-trained model that closely matches the use case requirements
- B
The time and resources required to collect and label a custom dataset
- C
The need for the model to adapt to highly domain-specific data
- D
The cost associated with hosting and serving the model on AWS
- E
The programming language used during model implementation
Show answer and explanation
Correct answers: A, B, C
Explanation
Choosing between a pre-trained model and a custom model involves evaluating factors such as the availability of pre-trained models for the use case, the effort required for creating a custom dataset, and the need for domain-specific adaptability. Pre-trained models are ideal when they align closely with the use case, while custom models are necessary for highly specialized tasks. Cost and programming language are secondary considerations that do not directly influence this decision.
- A. Correct.
If there is an open-source pre-trained model that closely matches the requirements, it may save time and effort compared to training a custom model.
- B. Correct.
Training a custom model requires significant time and resources to collect, prepare, and label a dataset, which should be considered in the decision-making process.
- C. Correct.
Highly domain-specific data often requires training a custom model, as pre-trained models may not generalize well to such specific use cases.
- D. Incorrect.
While cost is important, it is not directly related to the choice between pre-trained and custom models, as both can incur hosting and serving costs.
- E. Incorrect.
The programming language used is not a determining factor in choosing between pre-trained and custom models, as frameworks supporting both approaches work with multiple languages.