Demonstrate the Ability to Train and Inference CERTIFIED-AI-EXPERT Practice Questions (Page 1)

Questions 1–5

1. 1foundation · easy

   In the end-to-end training workflow for an AI model, which step typically comes immediately after data preparation?

   • AModel selection
   • BHyperparameter tuning
   • CModel deployment
   • DOutput interpretation
   Reveal Solution Discussion Select an answer first
2. 2application · medium

   A company has deployed a model for image classification. The model was trained on images of size 224x224. When a user uploads a 1000x1000 image, the system should:

   • ACrop the image to 224x224 without resizing.
   • BResize the image to 224x224 before passing it to the model.
   • CReject the image because it is not the correct size.
   • DPass the image directly to the model, as it can handle any size.
   Reveal Solution Discussion Select an answer first
4. 3foundation · easy

   In terms of latency requirements, how does inference typically differ from training?

   • ABoth require the same low latency
   • BInference often requires low latency for real-time predictions, while training has no strict latency requirement
   • CLatency is irrelevant for both training and inference
   • DTraining requires lower latency than inference
   Reveal Solution Discussion Select an answer first
5. 4application · medium

   A healthcare startup is training a model to classify medical images. They have a limited dataset and need to improve model accuracy. Which of the following actions is most aligned with the iterative training workflow?

   • ASwitch to a more complex model architecture without additional data.
   • BIncrease the number of epochs without changing the data or model architecture.
   • CApply data augmentation techniques and retrain the model with adjusted hyperparameters.
   • DDeploy the model immediately and collect feedback from users to improve accuracy.
   Reveal Solution Discussion Select an answer first
6. 5foundation · easy

   Why is model versioning an important consideration when deploying a model for inference?

   • ATo eliminate the need for monitoring
   • BTo reduce the latency of inference requests
   • CTo increase the model's accuracy automatically
   • DTo allow rollback to a previous version if the new model performs poorly
   Reveal Solution Discussion Select an answer first