A company needs to deploy a machine learning model for real-time predictions. Which approach minimizes latency?
ABatch prediction
BCloud Run
CDataflow with Pub/Sub
DVertex AI endpoints
Explanation
Vertex AI endpoints are designed for low-latency real-time predictions, unlike the other options that either introduce delays or do not fulfill the requirement.
Q432
What happens when you decrease the learning rate too much during training?
AFaster convergence
BRisk of overshooting minima
CSlower convergence
DNo effect on training
Explanation
A lower learning rate results in slower convergence, which can prevent the model from learning effectively.
Q433
Which service is best for real-time data processing on GCP?
ACloud Dataflow
BBigQuery
CCloud Storage
DCloud Pub/Sub
Explanation
Cloud Dataflow is designed for real-time data processing, while others serve different purposes like storage and query analytics.
Q434
A company needs to securely share data with external partners. Which Google Cloud feature should they use?
ACloud Identity
BData Catalog
CShared VPC
DIAM Roles
Explanation
IAM Roles provide fine-grained access control to share data securely.
Q435
What happens when using AutoML with imbalanced training data?
AMore accurate model predictions
BOverfitting on majority class
CUnderfitting on minority class
DImproved model interpretability
Explanation
Imbalanced training data can lead to the model overfitting on the majority class, poorly predicting the minority class.
Q436
Which service is best for training machine learning models at scale in Google Cloud?
AGoogle BigQuery
BGoogle Cloud AI Platform
CGoogle Cloud Functions
DGoogle App Engine
Explanation
Google Cloud AI Platform is optimized for ML model training, while others serve different purposes.
Q437
A company needs to analyze streaming data in real-time for decision making. Which service should they use?
AGoogle Cloud Pub/Sub
BGoogle Cloud Storage
CGoogle Cloud Spanner
DGoogle BigQuery
Explanation
Google Cloud Pub/Sub facilitates real-time data streaming, unlike the others which are suited for different tasks.
Q438
What happens when you set the model's learning rate too high?
AFaster convergence of the model
BStable learning without fluctuations
CPotential divergence of training
DImproved accuracy of predictions
Explanation
A high learning rate can cause divergence, making the model unable to learn effectively, while other options suggest desirable outcomes that are incorrect.
Q439
Which service is best for building scalable machine learning workflows?
ACloud Dataflow
BCloud Functions
CCloud Run
DCloud ML Engine
Explanation
Cloud ML Engine is designed for deploying ML models, while others serve different purposes.
Q440
A company needs to analyze large datasets but has strict cost limitations. Which Cloud service is most cost-efficient?
ABigQuery
BCloud SQL
CCloud Spanner
DCloud Dataproc
Explanation
BigQuery is designed for large queries with pay-per-query pricing, while others may incur higher fixed costs.