MLA-C01 Question 78
Select 3You are building a machine learning model to classify product images into different categories for an e-commerce platform. The model requires a high-quality labeled dataset to achieve accurate predictions. Which of the following AWS services or features can help you efficiently annotate and label your dataset while ensuring quality control?
- A
Amazon SageMaker Ground Truth with automated data labeling
- B
AWS Glue for data wrangling and schema transformations
- C
Amazon Rekognition for automated image analysis and labeling
- D
Amazon Mechanical Turk integration with SageMaker Ground Truth
- E
Human-in-the-loop (HITL) workflows in Amazon SageMaker
Show answer and explanation
Correct answers: A, D, E
Explanation
To create high-quality labeled datasets for your machine learning model, you can use Amazon SageMaker Ground Truth, which supports automated data labeling and human labelers via Amazon Mechanical Turk. Additionally, Human-in-the-loop (HITL) workflows in SageMaker further enhance quality control by combining machine learning with human judgment. These services are specifically designed for data annotation and labeling tasks, ensuring both efficiency and accuracy. AWS Glue and Amazon Rekognition, while useful for other purposes, are not directly applicable to this scenario.
- A. Correct.
Amazon SageMaker Ground Truth with automated data labeling is a key service for creating high-quality labeled datasets. It uses machine learning models to assist human labelers, making the process efficient and cost-effective.
- B. Incorrect.
AWS Glue is primarily used for data wrangling, schema transformations, and ETL tasks. It is not designed for annotating or labeling datasets.
- C. Incorrect.
Amazon Rekognition is used for image and video analysis, such as object detection and facial recognition. While it can analyze images, it is not specifically designed to label datasets for custom machine learning models.
- D. Correct.
Amazon Mechanical Turk integration with SageMaker Ground Truth allows you to leverage a large workforce of human labelers for annotation tasks, ensuring high-quality labels for your dataset.
- E. Correct.
Human-in-the-loop (HITL) workflows in Amazon SageMaker enable a combination of machine learning and human review for tasks that require human judgment, such as data labeling. This ensures quality control and accuracy in the labeled dataset.