Automating Model Retraining PROFESSIONAL-MACHINE-LEARNING-ENGINEER Practice Questions (Page 6)

Questions 26–30

1. 26foundation · easy

   Which of the following is a key component of a retraining policy?

   • AChoosing the programming language for the application
   • BDetermining the physical location of the data center
   • CSelecting the color palette for the model's UI
   • DDefining the criteria that trigger a retraining run
   Reveal Solution Discussion Select an answer first
2. 27foundation · easy

   What is a primary trade-off when choosing to retrain a model very frequently?

   • ALower operational cost but higher risk of data leakage
   • BHigher model freshness but increased compute and storage costs
   • CHigher model accuracy but slower inference speed
   • DLower model freshness but reduced operational complexity
   Reveal Solution Discussion Select an answer first
4. 28foundation · easy

   What is the primary purpose of continuous integration (CI) in an ML deployment pipeline?

   • ATo store training data in a versioned repository
   • BTo manually deploy the model to production after each change
   • CTo automatically build and test code changes before merging
   • DTo monitor model performance in production
   Reveal Solution Discussion Select an answer first
5. 29foundation · easy

   Which of the following is a valid trigger for initiating model retraining?

   • AA decrease in the cost of cloud storage
   • BA drop in the model's accuracy below a threshold
   • CA change in the color scheme of the application UI
   • DAn increase in the number of users accessing the model
   Reveal Solution Discussion Select an answer first
6. 30application · medium

   A team wants to use Cloud Build to automate the deployment of their ML model. They have a Docker image that contains the training code and a script to deploy the model to Vertex AI. What should they configure in Cloud Build to achieve this?

   • AA build trigger that runs the Docker image and executes the deployment script when a new commit is pushed to the repository.
   • BA Cloud Scheduler job that runs the Docker image on a schedule, but does not deploy the model.
   • CA Vertex AI pipeline that runs the training code and deploys the model, but does not use Cloud Build.
   • DA Cloud Function that triggers the deployment script when a new model is uploaded to Cloud Storage.
   Reveal Solution Discussion Select an answer first