Google Professional Machine Learning Engineer Question 111
Select 3Google Cloud PlatformYou are working as a machine learning engineer for a retail company that wants to build a product recommendation model using Vertex AI. You need to manage the dataset for training and ensure it is properly prepared for use in Vertex AI. Which of the following actions are necessary to manage the dataset in Vertex AI and ensure the model can be trained successfully?
- A
Upload the dataset to a Cloud Storage bucket and create a Dataset resource in Vertex AI.
- B
Ensure the dataset is in a format supported by Vertex AI, such as CSV or TFRecord.
- C
Manually preprocess the dataset by splitting it into training and validation sets before uploading it to Vertex AI.
- D
Define the schema of the dataset, including feature types and labels, when creating the Dataset resource in Vertex AI.
- E
Enable auto-splitting of the dataset into training, validation, and test sets in Vertex AI.
Show answer and explanation
Correct answers: A, B, D
Explanation
Managing datasets in Vertex AI involves uploading the dataset to a Cloud Storage bucket and creating a Dataset resource to make it accessible for training. It is also crucial to ensure the dataset is in a format supported by Vertex AI, such as CSV or TFRecord. Additionally, defining the schema of the dataset, including the feature types and labels, helps Vertex AI understand the data structure. While dataset splitting is useful, it is not mandatory to do this manually as Vertex AI allows configuration of splits during the training phase.
- A. Correct.
Correct: Uploading the dataset to a Cloud Storage bucket and creating a Dataset resource in Vertex AI is an essential step to manage datasets in Vertex AI.
- B. Correct.
Correct: Vertex AI requires datasets to be in specific formats like CSV or TFRecord, so ensuring the dataset is in a supported format is necessary.
- C. Incorrect.
Incorrect: While preprocessing is important, Vertex AI can handle dataset splitting automatically, so manual splitting is not strictly necessary.
- D. Correct.
Correct: Defining the schema of the dataset, such as specifying feature types and labels, is important to ensure Vertex AI can understand and process the data correctly.
- E. Incorrect.
Incorrect: Vertex AI does not have an explicit 'auto-splitting' feature for datasets. Splitting can be done manually or during the training configuration phase.