1.4 AI/Machine Learning 400-007 Practice Questions (Page 5)

Questions 21–25

1. 21foundation · easy

   Which of the following is an example of a business driver for adopting AI/ML?

   • AThe need to migrate all on-premises applications to the cloud
   • BThe availability of a new open-source machine learning framework
   • CA desire to increase the accuracy of demand forecasting to reduce inventory costs
   • DA requirement to increase the number of data scientists on staff
   Reveal Solution Discussion Select an answer first
2. 22foundation · easy

   What does data provenance refer to in the context of AI/ML data integrity?

   • AThe documented origin and lineage of data used in training and inference
   • BThe encryption method used to protect data at rest
   • CThe speed at which data can be ingested into the training pipeline
   • DThe format in which data is stored in the data lake
   Reveal Solution Discussion Select an answer first
4. 23application · medium

   A multinational bank must deploy an AI model for fraud detection that processes customer transaction data. The bank's legal team mandates that all personal data of EU residents must remain within the EU. The bank also requires low-latency inference for real-time transaction scoring. Which design approach best satisfies these constraints?

   • ATrain the model in a public cloud in the US and deploy inference in an EU region, copying only the model weights to the EU.
   • BDeploy the entire AI pipeline in a public cloud region located within the EU, using that region for both training and inference.
   • CUse a hybrid cloud with training in the EU and inference in a US region, relying on encryption to protect data in transit.
   • DUse a private cloud hosted in the bank's on-premises data center in the US, with data replicated to an EU location for compliance.
   Reveal Solution Discussion Select an answer first
5. 24foundation · easy

   Which activity is an example of an assurance mechanism for an AI/ML system?

   • APeriodically validating model predictions against known outcomes
   • BIncreasing the number of GPUs to speed up training
   • CBacking up the model files to a separate storage location
   • DSelecting a different algorithm for the model
   Reveal Solution Discussion Select an answer first
6. 25application · medium

   A media company is deploying a deep learning model for video content analysis. The training dataset is several terabytes and is stored in an on-premises data center. The company plans to train the model in the cloud. Which network design consideration is most important for the training phase?

   • AOptimize the inference path by placing the model close to the data source, ignoring the training data transfer.
   • BDeploy the training infrastructure on-premises to avoid data transfer altogether.
   • CUse a VPN over the public internet to transfer the data, as it is the most cost-effective option.
   • DEnsure high-bandwidth, low-latency connectivity between the on-premises data center and the cloud for efficient data transfer.
   Reveal Solution Discussion Select an answer first