Google Professional Machine Learning Engineer Question 170
Select 4Google Cloud PlatformYou are building a machine learning model to predict customer churn using TensorFlow in Vertex AI Workbench. You need to ensure that the development process is efficient, scalable, and integrates seamlessly with Google Cloud services. Which steps should you take to achieve this goal?
- A
Use a pre-configured Vertex AI Workbench JupyterLab environment with TensorFlow installed.
- B
Store the training and evaluation data in BigQuery and load it directly into your TensorFlow model using BigQuery APIs.
- C
Train the model locally on your Vertex AI Workbench instance without using any cloud-based resources.
- D
Use Vertex AI Training to run a distributed training job for the model using the TensorFlow framework.
- E
Export the trained model to TensorFlow SavedModel format and upload it to Cloud Storage for deployment.
Show answer and explanation
Correct answers: A, B, D, E
Explanation
Developing machine learning models in Vertex AI Workbench requires leveraging Google Cloud services to ensure scalability, efficiency, and seamless integration. Using the pre-configured environment, BigQuery for data handling, Vertex AI Training for distributed training, and exporting models to Cloud Storage for deployment are best practices that align with the recommended workflows for a machine learning engineer on Google Cloud.
- A. Correct.
Using a pre-configured Vertex AI Workbench JupyterLab environment simplifies the setup process, ensuring TensorFlow and other required libraries are available.
- B. Correct.
Storing data in BigQuery and using BigQuery APIs to load data into TensorFlow ensures scalability and seamless integration with Google Cloud services.
- C. Incorrect.
Training the model locally on the Workbench instance is not scalable for large datasets or complex models. Utilizing cloud-based resources is recommended for this scenario.
- D. Correct.
Using Vertex AI Training for distributed training allows you to handle large-scale models and datasets efficiently, leveraging cloud infrastructure.
- E. Correct.
Exporting the trained model to TensorFlow SavedModel format and uploading it to Cloud Storage enables easy deployment and compatibility with other GCP services.