Scribd
Upload
23 views21 pages

AI900 Openall

The document outlines the syllabus and key concepts for the Microsoft Azure AI Fundamentals AI-900 exam, covering topics such as AI workloads, machine learning, computer vision, and natural language processing. It emphasizes the importance of responsible AI practices, including fairness, reliability, and transparency. Additionally, it provides insights into Azure ML architecture, cognitive services, and the differences between various AI techniques like classification and object detection.

Uploaded by

navya sri
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
23 views21 pages

AI900 Openall

The document outlines the syllabus and key concepts for the Microsoft Azure AI Fundamentals AI-900 exam, covering topics such as AI workloads, machine learning, computer vision, and natural language processing. It emphasizes the importance of responsible AI practices, including fairness, reliability, and transparency. Additionally, it provides insights into Azure ML architecture, cognitive services, and the differences between various AI techniques like classification and object detection.

Uploaded by

navya sri
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 21

Microsoft Azure AI

Fundamentals
AI 900
Basics of Cloud and Microsoft Azure
Exam Syllabus and Agenda

• AI Workload and Consideration


• Machine Learning on Azure
Break
• Computer Vision
• Natural Language Processing
• Features of Conversational AI

• Praveen’s tip: Theoretical Exam, no practical experience or coding


experience or knowledge is required.
Basics Concepts
Machine Learning: Ability of the Machine to learn from the end
Artificial user or developer and perform the same action again.
Intelligence Artificial Intelligence: Produce what natural intelligent does via
Machine.
Machine Prediction : What can happen in the future is called prediction
Learning Forecasting: Prediction using previous event data
Anomaly Detection: Machine learn to find unexpected value or
events
Prediction • It works real time
and
Forecasting • Works based upon historical data
• Boundaries can be mutable
Anomaly • Can take preemptive action
Detection • Cannot predict when the problem will occur
• Analyze data over time : Yes
Azure AI
Responsible AI

Fairness : Treat everyone equally

Reliability and Safety: Consider the situation. Continuous Testing & Update

Privacy & Security: How the date is being handled?

Inclusiveness: everyone should get benefit

Transparency: how the AI system makes the decision?

Accountability: People who develop the system are accountable for governance.
Machine Learning

• Rule Based(Static) vs Machine Learning(Dynamic)


• Label and Features: What is to be predicted is Label and data which is used
for prediction is Features
• Thumb Rule: Minimum Redundancy and Maximum relevance
• Training Data vs Validation Data vs Testing Data
• Basic Algorithms
Feature +
• Classification Label
Algorithm Model
• Regression
• Cluster
• How to select Algorithm?: Based upon accuracy, Training Time, Lenearity,
Number of Predicted Columns, Number of Features
• Automated ML
• Azure ML Studio
Basic Algorithms..
Building model..
Classification Regression Clustering 1
Train data
Algorithm
Val data
Supervised learning Supervised learning Unsupervised
learning
2
Yes or No Numerical value, Grouping 3
continuous form
60% Model
Binary Grouping based
classification(2), upon features 4
Multiclass No label
classification(more 5
than 2) Test Data 90% Model
Automated ML Automated ML Not in Automated
ML 6

Final Model
Feature Selection and Feature Engineering..

• Label and Feature


Candidate
ID Skill Relevance Previously Employed? Total Number of Exp CTC Screen Select or Reject?
1Java Yes No 5 10No
2Python No Yes 10 20No
3Java Yes Yes 5 8Yes
4Cloud No No 5 10No
5Java Yes Yes 3 7Yes

Label
Features

• How it works?
ML
Features Labels
Algm

• Automatically or manually select those features which contribute most to your prediction
value.
Azure ML Architecture
Azure ML Portal Walkthrough

• Azure ML Studio Walk Through


• Creation of storage accounts
• Creation of dataset
• Creation of Instance
• Creation of Pipeline
• Azure Automated ML
Note: Very costly, be careful to delete after you try 
Let’s take a Break 
Microsoft Cognitive Services
Computer Vision Service

• Classification vs Object detection vs Sematic Segmentation


• Optical Character Recognition
• Face Detection(Detect human face), Recognition(Recognize the
detected human face) and Analysis
• Types of custom vision service:
• Extract info from video or image
• Analyze and Describe image
• Read text in image
• Read Handwriting
• Recognize Celebrities and Landmark
AI Demos (microsoft.com)
Classification vs Object Detection vs
Semantic Segmentation..
Image Classification Object Detection Semantic Segmentation
Detect primary object Find the object and their Pixel wise classification
within the image location

Will List of objects in the Return results in Pixel values


image boundary box

Single object Multiple Object Multiple Object


Optical Character Recognition

• Extract text from an image or handwritten notes.


• Uses Machine Learning algorithm to enhance the efficiency.
• Supports Rest API
• Demo: Vision Studio (azure.com)
Face Services
• Identify face in the image
• Face detection and Identification
• Emotion analysis
• face ID is a unique identifier string for each detected face in an image
• Not available for public.
Form Recognition Service

• AI based document processing solution


• Structured, Semi Structured and Unstructured Documents
• File size should be less than 500MB
• It can be Integerated with Power Automate
Custom vision service
• Build your own computer vision service
• Classification, Object Detection
• Multilabel(Multiple tags per image) vs Multiclass(Single tag per
image)
• Custom Vision – Home
Natural Language Processing
• Key phrase extraction: Extract the key words from bigger amount of
text.
• Entity Recognition: Returns the list of entities
• Sentiment Analysis: Understand the emotions of the the text
• Language detection: Understand the languge of the text (
Language Studio - Microsoft Azure)
• Converstion: Text to Speech, Speech to Text
• Translation Service: Translate from one language to another.
• Speech Service: Speech to Speech (AI Demos (microsoft.com))
• Language Understanding Service (Language Studio - Microsoft Azure)
Conversational AI

QnA Maker Chatbot

Knowledge Based Input based

No Learning Self Learning Capabilities

Static Dynamic

Standalone Can be Integerated with MS teams and


other channels.
Getting Decommissioned in 2025 Chatgpt

You might also like