MLS-C01 Question 321
Single answerYour company is building a machine learning solution to classify customer support tickets into predefined categories such as 'Billing', 'Technical Support', and 'General Inquiry'. The dataset contains thousands of labeled examples and follows a standard text classification structure. The team wishes to minimize the time spent on model development and deployment. How should you proceed?
- A
Use an Amazon SageMaker built-in algorithm for text classification, such as BlazingText, to quickly build and deploy the model.
- B
Build a custom model by training a deep learning model on Amazon SageMaker using TensorFlow or PyTorch to achieve higher accuracy.
- C
Use Amazon SageMaker Autopilot to automatically explore different machine learning models and automatically deploy the best-performing one.
- D
Train a custom deep learning model locally and then deploy it using Amazon SageMaker endpoints.
Show answer and explanation
Correct answer: A
Explanation
For standard machine learning tasks like text classification with labeled data and predefined categories, SageMaker's built-in algorithms are optimized to provide quick and accurate results. In this scenario, the BlazingText algorithm is well-suited for the task, saving time and effort while maintaining accuracy. Custom models or Autopilot could work, but they introduce unnecessary complexity given the straightforward nature of the problem.
- A. Correct.
This is the correct answer. Amazon SageMaker provides built-in algorithms like BlazingText that are specifically designed for text classification tasks. Using a built-in algorithm minimizes development time and is ideal for this standard use case.
- B. Incorrect.
This is not the best choice. While building a custom model using TensorFlow or PyTorch could achieve high accuracy, it requires significant time and expertise, which is unnecessary for this straightforward classification task.
- C. Incorrect.
Using SageMaker Autopilot is a valid option, but it may introduce unnecessary overhead compared to directly using a built-in algorithm like BlazingText for a standard classification use case.
- D. Incorrect.
Training a custom model locally and deploying it via SageMaker endpoints is not ideal in this scenario. It increases complexity and does not leverage the built-in capabilities of SageMaker for text classification.