FREE AI Engineer: Designing and Implementing an Azure [AI-100] Questions and Answers

Question 1

You must assess the trends in fuel prices over ten years. The solution must show visual representations of anomalous price swings.

Which Azure Cognitive Services API should you use?

Accepted Answer

Anomaly Detector

Answer

The Azure Anomaly Detector API is specifically designed to identify unusual patterns or outliers in time-series data, such as fuel prices over ten years. It can automatically detect various types of anomalies, including spikes, dips, and trend changes, making it ideal for pinpointing anomalous price swings and visualizing these deviations from normal behavior.

Question 2

You intend to analyze the medical records of patients spread out across the globe. You must suggest an alternative to using the cloud for data processing and storage.

Accepted Answer

An Apache Spark cluster that uses MMLSpark

Answer

An Apache Spark cluster, especially when enhanced with MMLSpark (Microsoft Machine Learning for Apache Spark), provides a powerful, scalable, and distributed framework for processing large datasets. By deploying this on-premises or in a private cloud environment, it offers an alternative to public cloud services for data processing and storage, which can be crucial for sensitive medical records that might have strict data residency or compliance requirements.

Question 3

Your business has a data center on-site. You want to release an app that uses the Face API to identify a group of people. The model has been honed. You must ensure that all image processing occurs in the on-site data center.

What should you deploy to host the Face API?

Accepted Answer

A Docker container

Answer

To ensure all image processing for the Face API occurs on-site, deploying the Face API as a Docker container is the correct solution. Azure Cognitive Services containers allow you to run specific Cognitive Services APIs, like the Face API, in your own on-premises environment or any other location, providing full control over data residency and processing location while still leveraging Azure's pre-trained models.

Question 4

A product catalog can be searched using the Bing Search service you have. You must locate the following details:
- The locale of the query
- The top 50 query strings
- The number of calls to the service
- The top geographical regions of the service
What should you implement?

Accepted Answer

Bing Statistics

Answer

Bing Statistics (or Bing Webmaster Tools analytics, which includes similar data) provides detailed insights into how users are interacting with Bing Search services. It offers metrics such as query locale, popular search queries, total calls to the service, and geographical distribution of users. This makes it the direct and most appropriate tool for gathering the specified usage and performance data for a Bing Search service.

Question 5

You have a real-time updating Face API solution. On a small dataset, the solution's pilot executes successfully. The performance deteriorates when you try to apply the method to a more extensive dataset constantly changing, lengthening the time it takes to identify existing faces. It would help if you suggested improvements to decrease the time it takes to recognize familiar faces without raising costs. What ought you offer?

Accepted Answer

Separate training into an independent pipeline and schedule the pipeline to run daily.

Answer

When dealing with a constantly changing, extensive dataset in a real-time Face API solution, performance deterioration often stems from inefficient model updates. Separating the training process into an independent pipeline allows the recognition system to use a stable, pre-trained model for inference, preventing real-time operations from being bogged down by continuous retraining. Scheduling daily training ensures the model remains current without impacting recognition speed or incurring excessive costs from constant, on-demand retraining.

AI - Engineer Practice Test

AI - Engineer Practice Test

FREE AI Engineer: Designing and Implementing an Azure [AI-100] Questions and Answers