Google Professional Machine Learning Engineer exam dumps

Google Professional Machine Learning Engineer practice question 258 of 522

Professional Machine Learning Engineer. Professional level, Google Cloud. Free question with the correct answer and a full explanation.

Google Professional Machine Learning Engineer Question 258

Select 2Google Cloud Platform

You are tasked with deploying a machine learning workflow for predicting customer churn using tabular data. The team has selected AutoML Tables for model training and evaluation. Once the model is trained and deployed, it needs to handle incoming prediction requests at scale. You are responsible for setting up an infrastructure that scales automatically based on demand. Which components should you use to ensure the solution meets these requirements?

  1. A

    Google Kubernetes Engine (GKE) to deploy a custom prediction service that scales automatically

  2. B

    AutoML Tables' built-in prediction service for handling prediction requests

  3. C

    Cloud Functions to host the AutoML Tables model

  4. D

    A GKE cluster with an autoscaler for training the AutoML Tables model

  5. E

    Cloud Pub/Sub to queue prediction requests before invoking the AutoML Tables model

  6. F

    AI Platform Prediction for scalable deployment of the trained AutoML Tables model

Show answer and explanation

Correct answers: B, F

Explanation

For this use case, AutoML Tables' built-in prediction service (Option 2) is the most straightforward and scalable choice to handle prediction requests. Additionally, AI Platform Prediction (Option 6) can also be used to deploy and scale trained models, including AutoML models, if more flexibility or advanced configuration is required. Other options involve unnecessary complexity or are not applicable to this specific scenario.

  • A. Incorrect.

    While GKE can be used to deploy custom prediction services, it is unnecessary in this scenario because AutoML Tables provides a built-in and scalable prediction service.

  • B. Correct.

    AutoML Tables includes a built-in prediction service that can handle prediction requests at scale, making it a suitable choice for deployment.

  • C. Incorrect.

    Cloud Functions is not designed to host and serve machine learning models directly, especially for computationally intensive tasks like ML predictions.

  • D. Incorrect.

    Training AutoML Tables models does not require GKE. AutoML Tables handles training on Google Cloud's managed infrastructure.

  • E. Incorrect.

    Cloud Pub/Sub is a messaging system and is not directly related to hosting or deploying ML models for real-time predictions.

  • F. Correct.

    AI Platform Prediction is another scalable option for deploying trained models, including those from AutoML Tables. However, AutoML Tables' built-in service is typically preferred for simplicity.

Timed practice exam

Take a Google Professional Machine Learning Engineer practice test under exam conditions

60 questions in 120 minutes, drawn from this bank, with a score report and a per-question review when you finish.

Start timed exam