AI, Machine Learning & Deep Learning

AI Concepts

• AI involves creating systems to perform tasks typically requiring human intelligence.
• Machine Learning (ML) is an AI subset enabling systems to learn from data without explicit programming.
• ML algorithms improve with increased data exposure.
• Deep Learning (DL) is an ML subfield using multi-layered artificial neural networks for data analysis.
• DL excels at image recognition and natural language processing.
• Computer Vision enables computers to "see" and interpret images.
• Natural Language Processing (NLP) focuses on enabling computers to understand and process human language

Machine Learning

• Supervised learning involves training a model on labeled data with known desired outputs.
• Common supervised learning tasks include classification and regression.
• Classification predicts a category or class label.
• Regression predicts a continuous numerical value.
• Unsupervised learning uses unlabeled data to discover patterns and relationships.
• Clustering and dimensionality reduction are common unsupervised learning tasks.
• Clustering groups similar data points.
• Dimensionality reduction reduces dataset variables while preserving key information.
• Reinforcement learning trains an agent to make decisions in an environment to maximize a reward.
• The agent learns through trial and error, with rewards or penalties as feedback.

Computer Vision

• Image classification assigns a label to an entire image.
• Object detection identifies and locates multiple objects within an image.
• Semantic segmentation classifies each pixel in an image.
• Image analysis applications include content moderation, Optical Character Recognition (OCR), and facial recognition.
• Video analysis applications include motion detection and object tracking.

Natural Language Processing

• Language models predict the probability of word sequences in a sentence.
• Common NLP tasks include sentiment analysis and key phrase extraction.
• Other NLP tasks: language detection, machine translation, question answering, speech recognition, and bot applications.

Responsible AI

• Responsible AI focuses on designing, developing, and deploying AI systems ethically and trustworthily.
• Key principles include fairness, ensuring no discrimination.
• Reliability and Safety: AI systems perform reliably and safely.
• Privacy and Security: Protecting sensitive data and ensuring AI systems' security.
• Inclusiveness: AI systems are accessible and beneficial to everyone.
• Transparency: Clear explanations of how AI systems work.
• Accountability: Clear lines of responsibility for AI systems.
• Microsoft's AI principles guide their Responsible AI approach.

Azure AI Services

• Azure AI services offer pre-built AI models and tools for developers to add AI capabilities to applications.
• Categories include Vision, Speech, Language, and Decision.
• Azure AI Vision extracts information from images and videos.
• Computer Vision analyzes images for objects, faces, text, and descriptions.
• Face service detects and identifies human faces in images.
• Azure AI Speech converts speech to text and text to speech.
• Speech to Text transcribes audio into text.
• Text to Speech converts text into natural-sounding speech.
• Azure AI Language provides NLP capabilities for understanding and analyzing text.
• Text Analytics extracts insights from text, such as sentiment, key phrases, and language used.
• Language Understanding (LUIS) builds conversational AI applications.
• QnA Maker creates knowledge bases from frequently asked questions and answers.
• Azure AI Decision helps in building AI systems that make better decisions.
• Anomaly Detector identifies anomalies in time series data.
• Content Moderator detects potentially offensive or unwanted content.
• Azure Machine Learning is a cloud platform for building, training, and deploying ML models.
• It provides a collaborative environment for data scientists and ML engineers.
• It supports Automated Machine Learning (AutoML), Azure Machine Learning Designer, Python SDK, R SDK, and CLI.

Azure Machine Learning

• Key components include workspaces for managing assets and resources.
• Datasets are abstractions for data used in ML experiments.
• Compute targets are infrastructure for running training and inference workloads.
• Pipelines are workflows defining ML task sequences.
• Models are trained ML models deployable for inference.
• Endpoints are HTTP endpoints for serving models.
• Automated Machine Learning (AutoML) automates building and training ML models.
• It selects the best algorithm and hyperparameters for a given dataset and task.
• Azure Machine Learning Designer offers a drag-and-drop interface for building ML pipelines.
• It allows creating and deploying ML models without coding.

Deploying and Managing Models

• Models can be deployed to Azure Kubernetes Service (AKS), Azure Container Instances (ACI), Azure Functions, and Azure Machine Learning compute.
• Azure Machine Learning provides tools for monitoring model performance and managing model versions.
• It allows tracking metrics like accuracy, precision, and recall.
• Azure Machine Learning supports model retraining to improve performance with new data.

Key Concepts Review

• AI is a broad field encompassing the creation of intelligent systems while machine learning and deep learning represent specialized subfields.
• Supervised, unsupervised, and reinforcement learning are the primary paradigms in machine learning, each suited to different types of problems and data.
• Computer vision empowers machines to interpret and understand images and videos, enabling a wide range of applications.
• Natural language processing focuses on enabling machines to understand, interpret, and generate human language.
• Responsible AI emphasizes the ethical and trustworthy development and deployment of AI systems, ensuring fairness, reliability, privacy, and transparency.
• Azure AI services provide a comprehensive suite of pre-built AI models and tools, enabling developers to easily add AI capabilities to their applications.
• Azure Machine Learning is a cloud-based platform for building, training, and deploying machine learning models, offering a collaborative environment and a variety of tools and frameworks.
• Automated Machine Learning (AutoML) automates the process of building and training ML models, simplifying the development process.
• Models can be deployed to various compute targets, and Azure Machine Learning provides tools for monitoring model performance and managing model versions.