Google Professional Machine Learning Engineer Question 451
Select 3Google Cloud PlatformYou are a Machine Learning Engineer at a retail company using Vertex AI for training and deploying models. Your team needs to ensure traceability and reproducibility of the machine learning pipeline, including tracking data transformations, model versions, and the associated metadata. Which features or practices in Google Cloud should you implement to achieve this?
- A
Enable Vertex ML Metadata to track artifacts, executions, and lineage.
- B
Use Cloud Logging to store a copy of all training datasets.
- C
Implement Vertex Pipelines to orchestrate and automatically log lineage of steps.
- D
Store model versions in Cloud Storage with a naming convention for versioning.
- E
Use BigQuery to query and analyze lineage metadata stored by Vertex AI.
Show answer and explanation
Correct answers: A, C, E
Explanation
To ensure traceability and reproducibility of machine learning workflows, Google Cloud provides tools such as Vertex ML Metadata, Vertex Pipelines, and BigQuery. Vertex ML Metadata tracks artifacts and lineage, Vertex Pipelines orchestrates workflows while logging lineage, and BigQuery allows for querying lineage metadata. Together, these tools provide an end-to-end solution for managing model and data lineage in a robust and scalable manner.
- A. Correct.
Correct. Vertex ML Metadata is specifically designed to track artifacts, executions, and lineage of machine learning workflows, enabling traceability.
- B. Incorrect.
Incorrect. While Cloud Logging is important for logging system events, it is not suitable for storing or tracking training datasets for lineage purposes.
- C. Correct.
Correct. Vertex Pipelines orchestrate ML workflows and automatically log lineage, including the steps taken during data transformations and model training.
- D. Incorrect.
Incorrect. While naming conventions for model versions in Cloud Storage can help with some level of organization, it does not provide detailed lineage or metadata tracking.
- E. Correct.
Correct. BigQuery can be used to analyze lineage metadata stored by Vertex AI, making it a valuable tool for understanding and querying the lineage data.