Google Professional Machine Learning Engineer Question 113
Select 3Google Cloud PlatformYou are a Machine Learning Engineer tasked with training a new image classification model using Vertex AI. Your dataset is stored in a Google Cloud Storage bucket. To ensure the dataset is managed effectively in Vertex AI for training, what steps should you take?
- A
Use the Vertex AI Dataset Import tool to load the data from the Cloud Storage bucket.
- B
Manually split the dataset into training, validation, and test sets before importing it into Vertex AI.
- C
Allow Vertex AI to automatically split the dataset into training, validation, and test sets during the model training pipeline.
- D
Ensure the dataset files are structured according to the requirements for Vertex AI dataset imports (e.g., organized into folders by label).
- E
Configure a Dataflow pipeline to preprocess and upload the data directly to Vertex AI's managed dataset storage.
Show answer and explanation
Correct answers: A, C, D
Explanation
When managing datasets in Vertex AI, it is important to use tools like the Dataset Import feature to load data from a Cloud Storage bucket, ensure the dataset is properly structured, and leverage Vertex AI's automatic data splitting capabilities. This streamlines the process and ensures compatibility with Vertex AI's training pipelines. Manual splitting or creating custom pipelines, like with Dataflow, may introduce unnecessary complexity in this scenario.
- A. Correct.
Correct: The Vertex AI Dataset Import tool can be used to load data from a Cloud Storage bucket into a managed dataset in Vertex AI. This is a key step in managing datasets in Vertex AI.
- B. Incorrect.
Incorrect: While it is possible to manually split the data, Vertex AI provides built-in options to handle this automatically during training, reducing manual effort.
- C. Correct.
Correct: Vertex AI can automatically handle dataset splitting into training, validation, and test sets during the model training pipeline if configured appropriately.
- D. Correct.
Correct: Vertex AI requires datasets to be structured properly (e.g., folders by label for image datasets) for successful import and model training.
- E. Incorrect.
Incorrect: Configuring a Dataflow pipeline is unnecessary for this scenario. While Dataflow can preprocess data, it is not required for managing datasets directly in Vertex AI.