Google Cloud

Google Cloud Certified – Professional Machine Learning Engineer

PR000269
Trending

Validate your skills as a Professional Machine Learning Engineer with exam code PR000269 in Google Cloud.

495 questions 0 views Free
Start Mock Test Timed · Full-length · Scored

Questions 381–390 of 495

Q381

You are configuring a machine learning model for image classification. What happens when you increase the number of hidden layers in your neural network?

  • A It reduces overfitting
  • B It may improve learning capacity
  • C It simplifies the model
  • D It guarantees better accuracy
Explanation Increasing hidden layers may improve learning capacity, but it doesn't guarantee better accuracy or prevent overfitting.
Q382

Which service is best for managing ML experiment tracking?

  • A AI Platform Experiments
  • B Cloud Functions
  • C BigQuery Machine Learning
  • D Cloud Storage
Explanation AI Platform Experiments is specifically built for tracking and managing ML experiments, unlike the other options.
Q383

A company needs to serve real-time predictions for an application. Which architecture pattern should they follow?

  • A Batch processing with Apache Spark
  • B Serverless architecture with Cloud Functions
  • C Streaming data with Pub/Sub and ML model
  • D Data warehouse and scheduled queries
Explanation Using Pub/Sub with real-time ML models effectively manages live data streams for predictions.
Q384

What happens when you enable AutoML training for excessive labeled data?

  • A System crashes from overload
  • B Increased training time expected
  • C Better model accuracy achieved
  • D Maximum data limit exceeded error
Explanation While more data can improve accuracy, it will also increase the training time significantly.
Q385

Which Google Cloud service is best suited for serverless machine learning inference?

  • A Cloud Functions
  • B Compute Engine
  • C Cloud Run
  • D AI Platform
Explanation Cloud Run allows serverless deployment of ML models, while the others require more management or are not designed for serverless inference.
Q386

A company needs to securely share data with a third-party using Google Cloud. Which service is the best fit?

  • A Cloud Storage Signed URLs
  • B BigQuery Data Transfer Service
  • C Cloud Pub/Sub
  • D FireStore
Explanation Signed URLs allow secure, temporary access to private Cloud Storage objects, unlike the other options.
Q387

What happens when you deploy a Cloud ML model with an outdated TensorFlow version?

  • A Deployment fails immediately.
  • B Model runs but may behave unpredictably.
  • C The model auto-updates to latest.
  • D Deployment is paused until fixed.
Explanation An outdated version may lead to unexpected behavior without failing, while the other options are incorrect assumptions about version handling.
Q388

Which service is best for automating ML workflows?

  • A AI Platform Pipelines
  • B Cloud Functions
  • C Cloud Run
  • D Cloud Composer
Explanation AI Platform Pipelines is designed for orchestrating machine learning workflows, whereas the others focus on serverless compute or task automation.
Q389

A company needs to store large amounts of unstructured data. Which Google Cloud service should they use?

  • A Cloud SQL
  • B Firestore
  • C BigQuery
  • D Cloud Storage
Explanation Cloud Storage is optimized for storing unstructured data, unlike the other options which are suited for structured data.
Q390

You are configuring a model versioning strategy. What happens when you deploy a new model version?

  • A The old version is deleted
  • B It auto-reverts to previous version
  • C The old version is still active
  • D Both versions can run concurrently
Explanation When deploying a new model version, both can run concurrently unless explicitly configured otherwise.