Microsoft Microsoft Certifications AI-100 Questions & Answers

• Question 31:

  You deploy an infrastructure for a big data workload.

  You need to run Azure HDInsight and Microsoft Machine Learning Server. You plan to set the RevoScaleR compute contexts to run rx function calls in parallel.

  What are three compute contexts that you can use for Machine Learning Server? Each correct answer presents a complete solution.

  NOTE: Each correct selection is worth one point.

  A. SQL

  B. Spark

  C. local parallel

  D. HBase

  E. local sequential

  Correct Answer: ABC

  Remote computing is available for specific data sources on selected platforms. The following tables document the supported combinations. RxInSqlServer, sqlserver: Remote compute context. Target server is a single database node (SQL

  Server 2016 R Services or SQL Server 2017 Machine Learning Services). Computation is parallel, but not distributed.

  RxSpark, spark: Remote compute context. Target is a Spark cluster on Hadoop.

  RxLocalParallel, localpar: Compute context is often used to enable controlled, distributed computations relying on instructions you provide rather than a built-in scheduler on Hadoop. You can use compute context for manual distributed

  computing.

  References:

  https://docs.microsoft.com/en-us/machine-learning-server/r/concept-what-is-compute-context

• Question 32:

  Your company has 1,000 AI developers who are responsible for provisioning environments in Azure.

  You need to control the type, size, and location of the resources that the developers can provision.

  What should you use?

  A. Azure Key Vault

  B. Azure service principals

  C. Azure managed identities

  D. Azure Security Center

  E. Azure Policy

  Correct Answer: B

  When an application needs access to deploy or configure resources through Azure Resource Manager in Azure Stack, you create a service principal, which is a credential for your application. You can then delegate only the necessary permissions to that service principal.

  References: https://docs.microsoft.com/en-us/azure/azure-stack/azure-stack-create-service-principals

• Question 33:

  You are designing an AI solution in Azure that will perform image classification.

  You need to identify which processing platform will provide you with the ability to update the logic over time. The solution must have the lowest latency for inferencing without having to batch.

  Which compute target should you identify?

  A. graphics processing units (GPUs)

  B. field-programmable gate arrays (FPGAs)

  C. central processing units (CPUs)

  D. application-specific integrated circuits (ASICs)

  Correct Answer: B

  FPGAs, such as those available on Azure, provide performance close to ASICs. They are also flexible and reconfigurable over time, to implement new logic. Incorrect Answers:

  D: ASICs are custom circuits, such as Google's TensorFlow Processor Units (TPU), provide the highest efficiency. They can't be reconfigured as your needs change.

  References: https://docs.microsoft.com/en-us/azure/machine-learning/service/concept-accelerate-with-fpgas

• Question 34:

  You have a solution that runs on a five-node Azure Kubernetes Service (AKS) cluster. The cluster uses an N-series virtual machine.

  An Azure Batch AI process runs once a day and rarely on demand.

  You need to recommend a solution to maintain the cluster configuration when the cluster is not in use. The solution must not incur any compute costs.

  What should you include in the recommendation?

  A. Downscale the cluster to one node

  B. Downscale the cluster to zero nodes

  C. Delete the cluster

  Correct Answer: A

  An AKS cluster has one or more nodes.

  References: https://docs.microsoft.com/en-us/azure/aks/concepts-clusters-workloads

• Question 35:

  Your company has recently deployed 5,000 Internet-connected sensors for a planned AI solution.

  You need to recommend a computing solution to perform a real-time analysis of the data generated by the sensors.

  Which computing solution should you recommend?

  A. an Azure HDInsight Storm cluster

  B. Azure Notification Hubs

  C. an Azure HDInsight Hadoop cluster

  D. an Azure HDInsight R cluster

  Correct Answer: A

• Question 36:

  You deploy an application that performs sentiment analysis on the data stored in Azure Cosmos DB.

  Recently, you loaded a large amount of data to the database. The data was for a customer named Contoso, Ltd.

  You discover that queries for the Contoso data are slow to complete, and the queries slow the entire application.

  You need to reduce the amount of time it takes for the queries to complete. The solution must minimize costs.

  What is the best way to achieve the goal? More than one answer choice may achieve the goal. Select the BEST answer.

  A. Change the request units.

  B. Change the partitioning strategy.

  C. Change the transaction isolation level.

  D. Migrate the data to the Cosmos DB database.

  Correct Answer: B

  References: https://docs.microsoft.com/en-us/azure/architecture/best-practices/data-partitioning

• Question 37:

  You have an AI application that uses keys in Azure Key Vault.

  Recently, a key used by the application was deleted accidentally and was unrecoverable.

  You need to ensure that if a key is deleted, it is retained in the key vault for 90 days.

  Which two features should you configure? Each correct answer presents part of the solution.

  NOTE: Each correct selection is worth one point.

  A. The expiration date on the keys

  B. Soft delete

  C. Purge protection

  D. Auditors

  E. The activation date on the keys

  Correct Answer: BC

  References: https://docs.microsoft.com/en-us/azure/architecture/best-practices/data-partitioning

• Question 38:

  You plan to implement a new data warehouse for a planned AI solution. You have the following information regarding the data warehouse:

  1.

  The data files will be available in one week.

  2.

  Most queries that will be executed against the data warehouse will be ad-hoc queries.

  3.

  The schemas of data files that will be loaded to the data warehouse will change often.

  4.

  One month after the planned implementation, the data warehouse will contain 15 TB of data.

  You need to recommend a database solution to support the planned implementation.

  What two solutions should you include in the recommendation? Each correct answer is a complete solution.

  NOTE: Each correct selection is worth one point.

  A. Apache Hadoop

  B. Apache Spark

  C. A Microsoft Azure SQL database

  D. An Azure virtual machine that runs Microsoft SQL Server

  Correct Answer: C

  References: https://docs.microsoft.com/en-us/azure/sql-database/saas-multitenantdb-adhoc-reporting

• Question 39:

  You need to build a solution to monitor Twitter. The solution must meet the following requirements:

  1.

  Send an email message to the marketing department when negative Twitter messages are detected.

  2.

  Run sentiment analysis on Twitter messages that mention specific tags.

  3.

  Use the least amount of custom code possible.

  Which two services should you include in the solution? Each correct answer presents part of the solution.

  NOTE: Each correct selection is worth one point.

  A. Azure Databricks

  B. Azure Stream Analytics

  C. Azure Functions

  D. Azure Cognitive Services

  E. Azure Logic Apps

  Correct Answer: BE

  References: https://docs.microsoft.com/en-us/azure/stream-analytics/streaming-technologies

  https://docs.microsoft.com/en-us/azure/stream-analytics/stream-analytics-twitter-sentiment-analysis-trends

• Question 40:

  You plan to build an application that will perform predictive analytics. Users will be able to consume the application data by using Microsoft Power BI or a custom website.

  You need to ensure that you can audit application usage.

  Which auditing solution should you use?

  A. Azure Storage Analytics

  B. Azure Application Insights

  C. Azure diagnostics logs

  D. Azure Active Directory (Azure AD) reporting

  Correct Answer: A