Describe the Solutions for Code, Data, and Model Traceability CERTIFIED-AI-EXPERT Practice Questions (Page 3)

Questions 11–15

1. 11foundation · easy

   What metadata is typically associated with a code commit in a version control system?

   • AAuthor, timestamp, and commit message
   • BModel accuracy and loss values
   • CCloud region and storage tier
   • DData schema and data quality scores
   Reveal Solution Discussion Select an answer first
2. 12expert · hard

   A global bank is deploying a loan approval model. Regulators require the bank to demonstrate that the model was trained on data that was processed in a compliant manner. The bank uses a data lake, a feature store, and a model registry, but the feature store does not record the version of the data used to create the features. What is the most effective way to ensure the audit trail is complete?

   • AStore the raw data files in the model registry alongside the model.
   • BUse the data lake's metadata to infer the data version used for each feature.
   • CEnhance the feature store to record the source data version and the transformation code version for each feature group.
   • DCreate a manual process to document the data version in a spreadsheet after each training run.
   Reveal Solution Discussion Select an answer first
4. 13foundation · easy

   What information is recorded for model traceability?

   • AOnly the final model file name
   • BModel version, training run ID, and hyperparameters
   • CThe cloud provider's billing ID
   • DThe number of GPUs used during training
   Reveal Solution Discussion Select an answer first
5. 14foundation · easy

   Which scenario benefits most from end-to-end traceability integration?

   • AOptimizing the number of training epochs
   • BChoosing a cloud region for model deployment
   • CAuditing which code and data versions produced a specific model
   • DReducing the size of the training dataset
   Reveal Solution Discussion Select an answer first
6. 15application · medium

   A logistics company's data science team is collaborating on a new routing optimization model. They need to ensure that all changes to the model training code are reviewed and that the final version used for training is clearly identified. What should they use?

   • AA code obfuscation tool to protect the code from unauthorized changes.
   • BA chat channel where team members announce when they have updated the code.
   • CA shared network drive where team members save their latest code changes.
   • DA Git branching strategy with pull requests for code review and tags for release versions.
   Reveal Solution Discussion Select an answer first