Artificial Intelligence Fundament

I have started the learning sessions provided in the Microsoft website. I thought it would be better to documentate the essentials of the studies for my own good.

Introduction

AI is a software which “mimics” a human being’s actions and functionalities, with such workloads:

• Machine Learning : Method of teaching a computer model to predict and make conclusions based on data
• Computer Vision
• Natural Language Processing
• Document Intelligence
• Knowledge Mining
• Generative AI

Machine Learning

How do machiens learn? They learn from data, which is abundant in the world that we live in today. Data scientists utilize these data in order to train machine learning models capable of making predictions. For this to work out, models must recognize the relationship between data. (Example: botanitsts and scientists develop an algorithm/pattern of classifying flowers as they study several of them)

Azure Machine Learning Studio provides these AI environments:

• Automated Machine Learning
• Azure Machine Learning : GUI-based machine learning solution
• Data Matrix Visualization
• Notebooks

Computer Vision

Computer vision is responsible for the “visual” aspect of AI. It involves various fundamental tasks:

• Image Classification
• Object Detection
• Semantic System Distinction : Each pixel gets classified based on the object that they belong in
• Image Analysis
• Face Detection, Analysis, and Identification
• Optical Character Recognition

Azure Vision Studio provides these CV environments:

• Image Analysis
• Face
• OCR

Natural Language Processing

NLP is the area of AI which is responsible for handling written and spoken language. Some examples of a NLP-based usages are deciphering emails, creating responses for audio input, translation, and understanding NL input.

Azure AI Language provides these AI environments:

• Azure AI Voice
• Azure AI Translator
• Azure Language Studo

Document Intelligence

Document Intelligence is an area of AI which takes care of mass management, processing, and utilization of data within documents and forms.

Azure AI Document Intelligence accelerates the process of accumulating data from scanned documents. This process, often referred to as “Knowledge Mining”, extracts a large quantity of irregular data to make a perfectly usable information centers.

Generative AI

Generative AI is capable of creating a seemingly original content. It accepts natural language input and returns natural language, image, code, or aurio-based output in an appropriate format.

Azure OpenAI Service is Microsoft’s cloud solution for deploying, hosting and configuring our very own generative AI.

Challenges involved in the implementation of AI

• Viruses can affect the result.
• Errors can bring deterimental effects.
• Data Exposure
• Solution may not be applicable to everyone
• Users must trust in the complicated system
• Whose responsibility is an AI-based choice?

Undertanding AI

Microsoft’s AI Software development adheres to six fundaments in order to prevent any negative outputs.

• Fairness : an AI system must treat everyone equally
• Fidelity and Stability : an AI system must work in a safe and reliable manner
• Privacy and Security : an AI system must protect private information
• Inclusiveness : an AI system must strengthen and build relationship among people
• Transparency : an AI system must be comprehensible and predictable
• Reponsibility : an AI system must be liable

Reference

https://learn.microsoft.com/ko-kr/training/modules/get-started-ai-fundamentals/7-challenges-with-ai