The requirement is to deploy a customized machine learning (ML) model and monitor its quality for potential drift over time in a production environment. Let’s evaluate each option:
A. Amazon Rekognition: This service is designed for image and video analysis, such as object detection, facial recognition, and text extraction. It is not suited for deploying custom ML models or monitoring model quality drift.
B. Amazon SageMaker Clarify: This feature helps detect bias in ML models and explains model predictions. While it addresses fairness and interpretability, it does not specifically focus on monitoring model quality drift over time in production.
C. Amazon Comprehend: This is a natural language processing (NLP) service for extracting insights from text, such as sentiment analysis or entity recognition. It does not support deploying custom ML models or monitoring model performance drift.
D. Amazon SageMaker Model Monitor: This feature is part of Amazon SageMaker and is specifically designed to monitor ML models in production. It tracks metrics such as data drift, model drift, and performance degradation over time, alerting users when issues are detected.
Exact Extract Reference: According to the AWS documentation on Amazon SageMaker, “Amazon SageMaker Model Monitor allows you to detect and remediate data and model quality issues in production. It continuously monitors the performance of deployed models, capturing data and model predictions to detect deviations from expected behavior, such as data drift or model performance degradation.” (Source: AWS SageMaker Documentation - Model Monitoring, https://docs.aws.amazon.com/sagemaker/latest/dg/model-monitor.html).
This directly aligns with the requirement to observe model quality drift, making Amazon SageMaker Model Monitor the correct choice.
[References:, AWS SageMaker Documentation: Model Monitoring (https://docs.aws.amazon.com/sagemaker/latest/dg/model-monitor.html), AWS AI Practitioner Study Guide (conceptual alignment with monitoring deployed ML models), , , ]
