AI-900 Primer (Azure AI Fundamentals) : Tips for AI-900 based on Exam given on 01/10/2021
Cleared AI-900 (on 01/10/2021) so thought of sharing my perspective before it slips away my unretentive mind.
60 Mins — 53 Questions
Well, during the initial look, it might appear that you have to quickly complete the questions since you have just 1 minute for each question, however, do not get overwhelmed as the questions are easy , as long as you are aware of Cognitive Services provided by Azure and have some working experience of building AI models on Azure. Here is the breakdown of the Topics covered in the test I gave -
- Natural Language Processing (NLP) Services on Azure (Including Conversational AI) — 50%
- Azure Responsible AI — 15%
- Computer Vision Services on Azure — 20 %
- General AI Concepts — 15%
Natural Language Processing (NLP) Services on Azure
Some of the most common concepts covered as part of the exam were topics like Text Analytics service, Speech Synthesis, Speech recognition etc.. There are subtle differences between the use-cases of these services which sometimes could be confusing so I highly recommend to read about these and understand the underlying purpose of each of these. Here is some information in brief -
What is Speech recognition?
The ability to detect and interpret spoken input.
Speech recognition is concerned with taking the spoken word and converting it into data that can be processed — often by transcribing it into a text representation. The spoken words can be in the form of a recorded voice in an audio file, or live audio from a microphone.
- Providing closed captions for recorded or live videos
- Creating a transcript of a phone call or meeting
- Automated note dictation
- Determining intended user input for further processing
What is Speech synthesis?
The ability to generate spoken output.
Speech synthesis is in many respects the reverse of speech recognition. It is concerned with vocalizing data, usually by converting text to speech
- Generating spoken responses to user input.
- Creating voice menus for telephone systems.
- Reading email or text messages aloud in hands-free scenarios.
- Broadcasting announcements in public locations, such as railway stations or airports.
What is Text Analytics?
Text analytics is a process where an artificial intelligence (AI) algorithm, running on a computer, evaluates these same attributes in text, to determine specific insights.
Some of the use-cases of Azure Text Analytics Service are:
· Determining the language of a document or text (for example, French or English)
· performs sentiment analysis on text to determine a positive or negative sentiment
· extracts key phrases from the text that might indicate its main talking points
· Identify and categorize entities in the text. Entities can be people, places, organizations, or even everyday items such as dates, times, quantities, and so on
What is the service from Microsoft Azure for Text Translation?
The Translator Text service, which supports text-to-text translation
What is the service from Microsoft Azure for Speech Translation?
The Speech service, which enables speech-to-text and speech-to-speech translation.
What is Language Understanding Intelligent Service (LUIS)?
Language Understanding (LUIS) is a cloud-based conversational AI service that applies custom machine-learning intelligence to a user’s conversational, natural language text to predict overall meaning, and pull out relevant, detailed information.
A client application for LUIS is any conversational application that communicates with a user in natural language to complete a task. Examples of client applications include social media apps, AI chatbots, and speech-enabled desktop applications.
Refer this cheat-sheet for a quick introduction to various Azure cognitive Services –
Conversational AI Services on Azure
The questions for Conversational AI were primarily focused around web chat bots, QnA maker and Knowledge base. Couple of questions that confused me a bit –
What are the different options to create a repository of questions and answers for chat bot?
· Generated from an existing FAQ document or web page.
· Imported from a pre-defined chit-chat data source.
· Entered and edited manually.
What are various ways to get response from chat-bot?
I do know that we can interact with a bot through web app and Microsoft Teams. Though I wasn’t sure if we can interact with a bot through email. While I know Cortana can send email confirmation for setting up meetings, but I haven’t tried responding back so it is worth confirming this once (exercise for you 😊).
What do you need to implement a conversation AI-based chatbot?
A knowledge base of question and answer pairs — usually with some built-in natural language processing model to enable questions that can be phrased in multiple ways to be understood with the same semantic meaning.
A bot service that provides an interface to the knowledge base through one or more channels.
Which Azure service is used to create and publish a knowledge base with built-in natural language processing capabilities?
Which Azure service provides a framework for developing, publishing, and managing bots on Azure?
Azure Bot Service.
Azure Responsible AI
Questions pertaining to Responsible AI were around ethical aspect of AI and its governance. Few questions were around identifying which Responsible AI criterion is fulfilled by a scenario while some other were related to the definition.
One question, I remember distinctly, was very confusing due to definition subtleties between “Accountability” and “Reliability & Safety”. As far as I can remember, it was identifying the correct principle for the following definitions –
“Process to ensure AI model decisions can b overridden manually” : I answered “Accountability”
“Building AI solutions to respond to unanticipated situations and avoiding harmful manipulations” : I answered “Reliability & Safety”
Here is a brief primer of Reliable AI principles-
What are six guiding principles of Reliable AI ?
AI systems should treat all people fairly. For example, suppose you create a machine learning model to support a loan approval application for a bank. The model should make predictions of whether or not the loan should be approved without incorporating any bias based on gender, ethnicity, or other factors that might result in an unfair advantage or disadvantage to specific groups of applicants.
Reliability and safety
AI systems should perform reliably and safely. For example, consider an AI-based software system for an autonomous vehicle; or a machine learning model that diagnoses patient symptoms and recommends prescriptions. Unreliability in these kinds of system can result in substantial risk to human life.
Privacy and security
AI systems should be secure and respect privacy. The machine learning models on which AI systems are based rely on large volumes of data, which may contain personal details that must be kept private. Even after the models are trained and the system is in production, it uses new data to make predictions or take action that may be subject to privacy or security concerns.
AI systems should empower everyone and engage people. AI should bring benefits to all parts of society, regardless of physical ability, gender, sexual orientation, ethnicity, or other factors.
AI systems should be understandable. Users should be made fully aware of the purpose of the system, how it works, and what limitations may be expected.
People should be accountable for AI systems. Designers and developers of AI-based solution should work within a framework of governance and organizational principles that ensure the solution meets ethical and legal standards that are clearly defined.
Computer Vision Services on Azure
This was very straightforward, and many questions were around Object detection, OCR, Face API, Read API. There were couple of questions around Custom vision models as well. Here are some good questions to read before the exam –
What are some examples of specialized domain models supported by Computer Vision service?
Celebrities — The service includes a model that has been trained to identify thousands of well-known celebrities from the worlds of sports, entertainment, and business.
Landmarks — The service can identify famous landmarks, such as the Taj Mahal and the Statue of Liberty.
What is the difference between Object detection and Image classification?
Image classification is a machine learning based form of computer vision in which a model is trained to categorize images based on the primary subject matter they contain.
Object detection goes further than this to classify individual objects within the image, and to return the coordinates of a bounding box that indicates the object’s location.
Which services are used to detect and analyze faces from Microsoft Azure?
Computer Vision, which offers face detection and some basic face analysis, such as determining age.
Video Indexer, which you can use to detect and identify faces in a video.
Face, which offers pre-built algorithms that can detect, recognize, and analyze faces.
General AI Concepts
There were quite a few questions around differentiating Classifications, Clustering and Regression so make sure you understand these concepts.
You must also read about various Model evaluation methods primarily for Regression (RMSE, R2) and Classification (Accuracy, Precision, Recall).
Few other things to know are –
What information is required by client application to call a ML model through API?
A key that is used to authenticate client applications.
An endpoint that provides the HTTP address at which your resource can be accessed.
What is a Compute Cluster vs Inference Cluster?
Scalable clusters of virtual machines for on-demand processing of experiment code.
Deployment targets for predictive services that use your trained models.
If you prep these concepts really well , I have no doubt that you can score really well in the exam. Do share your results if this article helps you towards getting AI-900 certified and will be glad to see that my notes have been helpful.
All the Best !!!