Transcript for:
Understanding Azure AI Fundamentals

Title: AI-900 Azure AI Fundamentals URL Source: blob://pdf/3919e69d-705a-47a8-a156-fa1a9f6bbbea Markdown Content: Copyright Microsoft Corporation. All rights reserved. Artificial Intelligence # An Introduction to # Cloud -based AI Dr Amir Pourabdollah Outline Cloud and cloud services Why Cloud -based AI? Common Workloads and Examples of Cloud -based AI Challenges and Risks of AI Principles of Responsible AI Sources: Microsoft Azure AI Fundamentals (AI -900) https://learn.microsoft.com/en -us/credentials/certifications/exams/ai -900/ Microsoft Azure Fundamentals: Describe cloud concepts https://learn.microsoft.com/en -us/training/paths/microsoft -azure -fundamentals -describe -cloud -concepts/ Microsoft Corporation > Azure What is cloud computing? Compute Storage Networking Analytics Cloud Computing is the delivery of computing services over the internet, enabling faster innovation, flexible resources, and economies of scale. Copyright Microsoft Corporation. All rights reserved. Cloud Benefits High availability Scalability Global reach Agility Disaster recovery Fault tolerance Elasticity Customer latency capabilities Predictive cost considerations Security Copyright Microsoft Corporation. All rights reserved. Cloud Challenges Connectivity Data Privacy Black -boxed Solution Costs (if not proportional to the service) Data Security Concerns QoS, Technology familiarity, Copyright Microsoft Corporation. All rights reserved. Public cloud Owned by cloud services or hosting provider. Provides resources and services to multiple organizations and users. Accessed via secure network connection (typically over the internet). Copyright Microsoft Corporation. All rights reserved. Private cloud Organizations create a cloud environment in their datacenter. Organization is responsible for operating the services they provide. Does not provide access to users outside of the organization. Microsoft Corporation > Azure Hybrid cloud Combines Public and Private clouds to allow applications to run in the most appropriate location. Copyright Microsoft Corporation. All rights reserved. Infrastructure as a Service (IaaS) Build pay -as -you -go IT infrastructure by renting servers, virtual machines, storage, networks, and operating systems from a cloud provider. Copyright Microsoft Corporation. All rights reserved. Platform as a Service (PaaS) Provides environment for building, testing, and deploying software applications; without focusing on managing underlying infrastructure. Microsoft Corporation > Azure Software as a Service (SaaS) Users connect to and use cloud -based apps over the internet: for example, Microsoft Office 365, email, and calendars. Microsoft Corporation > Azure Why Cloud -based AI All benefits of cloud technology, plus: Merging of ML/AI with cloud -based computing environments Opportunities for sharing knowledge and resources Offloading the computation power from client -side devices The ultimate solution with growing IoT, embedded AI, Do not reinvent the wheel! Microsoft Corporation > Azure Common Cloud Service Providers for AI Many comparison reviews, e.g., https://www.datamation.com/artificial - intelligence/the -top -cloud -based -ai -services/ Machine Learning Predictive models based on data and statistics the foundation for AI Anomaly Detection Systems that detect unusual patterns or events, enabling pre -emptive action Computer Vision Applications that interpret visual input from cameras, images, or videos Natural Language Processing Applications that can interpret written or spoken language, and engage in dialogs with human users Knowledge Mining Extract information from data sources to create a searchable knowledge store Generative AI Create contents using large trained/pre -trained models ## Common Workloads of Cloud -based AI Examples of Cloud -based Computer Vision Services Taxi Image Classification bus car Cyclist Cyclist bus Object Detection > bus > car > cyclist Semantic Segmentation A person with a dog on a street Image Analysis Face Detection & Recognition The Toronto Dominion Bank Optical Character Recognition Example Cloud Service: Anomaly Detection > Anomaly detection is the act of identifying events, or observations, that differ in a significant way from the rest of the data being evaluated. Anomalies can be detected by learning from the data patterns. Accurate anomaly detection leads to prompt troubleshooting. > The Anomaly Detector service can be used to create a cloud -based monitoring application Machine Learning on the Cloud Traditional machine learning model development is resource -intensive, requiring significant domain knowledge and time to produce and compare dozens of models. Using the cloud resources for computing -intensive tasks - Training models - Model optimisation - Running models Different approaches - Use pre -trained model - Visual interfaces for model training and deployment - Server -side programming - Hybrid (a combination of the above) > Data Model Example: Azure ML (reading only) Repeating experiments allows choosing the best model to fit data according to a chosen performance metric (supervised learning) Example: Azure ML cont. (reading only) Once the model is deployed, an endpoint and a key is allocated to the deployed model. These can be used (or shared with the users) to be used within client -side programs Predictions can be the results of repeated call to the deployed cloud service (without using the local resources) Example: Pipelining in Azure ML (reading only) Visual tool for creating a machine learning pipelines > Many pre -programmed modules > + support for user -programmed > modules Used for supervised and unsupervised methods Use a training pipeline to train and evaluate a model Create an inference pipeline to predict labels from new data (not shown) Deploy and publish the inference pipeline as a service for apps to use. NLP Capabilities on the Cloud Language Language detection Key phrase extraction Entity detection Sentiment analysis Question answering Conversational language understanding Speech Text to speech Speech to text Speech translation Translator Text translation Bot Service Platform for conversational AI Example: Developing a Cloud -based Chatbot (reading only) Define a knowledge base of question and answer pairs: By associating answers to [alternative] questions From an existing FAQ document By using built -in chit -chat (or a combination of the above) Train a model (if needed for answering unseen questions) Deploy as a Cloud service Consume the knowledge base from client apps, connect to the required channels Lab this week To use Azure AI service for image/ object recognition Challenges and Risks with AI Challenge or Risk Example Bias can affect results A loan -approval model discriminates by gender due to bias in the data with which it was trained Errors may cause harm An autonomous vehicle experiences a system failure and causes a collision Data could be exposed A medical diagnostic bot is trained using sensitive patient data, which is stored insecurely Solutions may not work for everyone A predictive app provides no audio output for visually impaired users Users must trust a complex system An AI -based financial tool makes investment recommendations - what are they based on? Who's liable for AI -driven decisions? An innocent person is convicted of a crime based on evidence from facial recognition who's responsible? Principles of Responsible AI Fairness Reliability & Safety Privacy & Security Inclusiveness Transparency Accountability > See resources and examples: https://www.microsoft.com/en -us/ai/principles -and -approach#resources ## Fairness 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 a 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. Application development must be subjected to rigorous test and deployment management processes to ensure that they work as expected before release. 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 s ubject to privacy or security concerns. Inclusiveness AI systems should empower everyone and engage people. AI should bring benefits to all parts of society, regardless of physic al ability, gender, sexual orientation, ethnicity, or other factors. Transparency AI systems should be understandable. Users should be made fully aware of the purpose of the system, h ow it works, and what limitations may be expected. Accountability People should be accountable for AI systems. Designers and developers should work within a framework of governance and organizational principles that ensure the solution meets ethical and legal standards that are clearly defined.