A company needs an ML model for image classification. Which approach is most appropriate?
AUse AutoML Vision
BBuild from scratch
CUse Cloud Run
DEmploy Dataflow
Explanation
AutoML Vision is tailored for image classification, unlike the others.
Q222
You are configuring a model on AI Platform. What does enabling GPU acceleration do?
AIncreases data redundancy
BSpeeds up model training
CDecreases cost
DEnsures data quality
Explanation
GPU acceleration primarily speeds up model training processes.
Q223
Which Google Cloud service is best for managing large-scale machine learning workflows?
ACloud AI Platform
BCloud Functions
CCloud Storage
DCompute Engine
Explanation
Cloud AI Platform is designed specifically for managing ML workflows, unlike the other options which focus on different functionalities.
Q224
A company needs to deploy a machine learning model but wants to minimize API latency. What should they choose?
ACloud Run
BCloud Functions
CAI Platform Predictions
DCloud Storage
Explanation
AI Platform Predictions is optimized for low-latency serving of models, while others are not primarily designed for this purpose.
Q225
What happens when you use an unbounded dataset in a BigQuery machine learning model?
AIt fails with error.
BIt works but uses all quota.
CTraining is inefficient.
DIt produces biased results.
Explanation
While training on unbounded datasets may succeed, it can lead to inefficiencies in resource usage and higher costs.
Q226
Which Google Cloud service is designed for deploying containerized applications?
AApp Engine
BCloud Functions
CGoogle Kubernetes Engine
DCloud Run
Explanation
Google Kubernetes Engine is specifically designed for managing container orchestration, unlike the others which serve different purposes.
Q227
A company needs to process data in real-time with Google Cloud. Which service is best suited for this requirement?
ABigQuery
BCloud Dataflow
CCloud Storage
DCloud Pub/Sub
Explanation
Cloud Dataflow is optimized for real-time data processing, while the others focus on different functionalities.
Q228
You are configuring your machine learning model for calibration. What technique would you apply to ensure the predicted probabilities are representative?
AOverfitting
BPlatt Scaling
CRegularization
DFeature Selection
Explanation
Platt Scaling is the standard technique used for probability calibration, while the other options serve different purposes in model training.
Q229
Which service is best for real-time data processing?
ACloud Pub/Sub
BCloud Functions
CCloud Storage
DBigQuery
Explanation
Cloud Pub/Sub is designed for real-time messaging, while the others are for batch processing or storage.
Q230
A company needs to securely share access keys. What service should they use?
ACloud Identity
BSecret Manager
CCloud IAM
DCloud Storage
Explanation
Secret Manager securely stores, manages, and accesses API keys.