Automated Retraining MACHINE-LEARNING-PROFESSIONAL Practice Questions (Page 2)

Questions 6–10

1. 6foundation · easy

   When comparing multiple retrained models, what is the primary basis for selecting the top-performing model?

   • AValidation metrics and business criteria
   • BThe model that uses the most training data
   • CThe model that trains the fastest
   • DThe model with the most complex architecture
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   A machine learning team wants to automatically retrain a model when its accuracy on live data falls below a predefined threshold. Which monitoring approach should they implement?

   • AData drift detection that monitors feature distributions
   • BLog monitoring that checks for application errors
   • CInfrastructure monitoring that tracks CPU and memory usage
   • DPerformance degradation monitoring that tracks live accuracy and triggers retraining when the threshold is breached
   Reveal Solution Discussion Select an answer first
4. 8expert · hard

   A company has an automated retraining pipeline that registers models in the MLflow Model Registry. The team wants to implement a champion-challenger approach where the challenger model is evaluated against the champion in a shadow deployment (i.e., running in parallel with the champion but not serving live traffic) before being promoted to production. What is the best way to implement this?

   • ARegister the challenger and immediately transition it to Production, then monitor its performance.
   • BRegister the challenger in the registry, deploy it to a shadow endpoint, collect its predictions and compare them to the champion's predictions on the same live traffic, and then promote it if it performs better.
   • CUse the MLflow Registry to compare the challenger's metrics to the champion's metrics, and promote if better.
   • DEvaluate the challenger offline on a historical dataset and promote it if it has better validation metrics.
   Reveal Solution Discussion Select an answer first
5. 9foundation · easy

   In a champion-challenger evaluation, what is the role of the 'champion' model?

   • AIt is the current model in production that serves as the baseline for comparison
   • BIt is a model used only for testing purposes
   • CIt is a newly trained model that has not yet been deployed
   • DIt is a model that has been retired from production
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   What is the purpose of a champion-challenger approach in model evaluation?

   • ATo select the model with the lowest inference latency
   • BTo train multiple models simultaneously on different hardware
   • CTo compare a newly trained model (challenger) against the current production model (champion) before deciding whether to replace it
   • DTo ensure the model is deployed to multiple regions
   Reveal Solution Discussion Select an answer first