AI and Machine Learning Flashcards

Flashcards for reviewing AI and Machine Learning principles, covering topics such as AI's definition, common workloads, responsible AI, machine learning types, regression, classification, clustering, Azure Machine Learning, computer vision, natural language processing, and generative AI.

Artificial Intelligence (AI)

Software that imitates human capabilities, including predicting outcomes, recognizing patterns, interpreting visual input, understanding language, and extracting information to gain knowledge.

Machine Learning

Predictive models based on data and statistics, forming the foundation for AI.

Computer Vision

Capabilities within AI used to visually interpret the world through cameras, video, and images.

Natural Language Processing (NLP)

Capabilities within AI for a computer to interpret written or spoken language and respond appropriately.

Document Intelligence

Capabilities within AI that manage, process, and use high volumes of data found in forms and documents.

Knowledge Mining

Capabilities within AI to extract information from large volumes of often unstructured data to create a searchable knowledge store.

Generative AI

Capabilities within AI that create original content in a variety of formats, including natural language, images, and code.

Fairness (Responsible AI)

AI systems should treat all people fairly, avoiding bias in predictions or decisions.

Reliability & Safety (Responsible AI)

AI systems should perform reliably and safely, minimizing risks to human life.

Privacy & Security (Responsible AI)

AI systems should be secure and respect privacy, protecting personal details.

Inclusiveness (Responsible AI)

AI systems should empower everyone and engage people, benefiting all parts of society.

Transparency (Responsible AI)

AI systems should be understandable, making users aware of the system's purpose, how it works, and its limitations.

Accountability (Responsible AI)

People should be accountable for AI systems, adhering to ethical and legal standards.

Supervised Machine Learning

Training data includes known labels.

Unsupervised Machine Learning

Training data is unlabeled.

Regression (Machine Learning)

Label is a numeric value; predicting a number based on input features.

Classification (Machine Learning)

Label is a categorization or class.

Clustering (Machine Learning)

Similar items are grouped together based on their features.

MAE

Mean absolute error; measures how close the predictions are to the actual values.

RSME

Root mean squared error; the square root of the average squared distance between the actual and the predicted values.

Binary Classification

Label is or is not a class.

Multiclass classification

Label is one of multiple classes.

Accuracy

The number of true positives and true negatives divided by the total number of predictions; the total of correct predictions.

Precision

The number of true positives divided by the sum of the number of true positives and false positives.

Recall

The number of true positives divided by the sum of the number of true positives and false negatives.

F-score

Combines precision and recall as a weighted mean value.

Area Under Curve (AUC)

A measure of true positive rate over true negative rate.

Azure Machine Learning

A cloud-based platform for machine learning, providing scalable resources for training, deploying, and managing models.

Compute Instance

Development workstations that data scientists can use to work with data and models.

Compute Clusters

Scalable clusters of virtual machines for on-demand processing of experiment code.

Inference Clusters

Deployment targets for predictive services that use your trained models.

Attached Compute

Links to existing Azure compute resources, such as Virtual Machines or Azure Databricks clusters.

Automated Machine Learning

A feature in Azure Machine Learning that simplifies model training for regression or classification.

Normalization

Mitigating possible bias by normalizing the numeric features, so they are on the same numeric scale.

Training Dataset

The sample of data used to train the model.

Validation Dataset

A sample of data used to validate the model and see if the model can correctly predict or classify.

Pipelines (Azure Machine Learning)

Multi-step workflows to prepare data, train models, and perform model management tasks in Azure Machine Learning.

Azure AI Services

A suite of services covering Vision, Speech, Language, Decision, and Generative AI.

Azure AI Search

Data extraction, enrichment, and indexing for intelligent search and knowledge mining.

Computer Vision

Creating solutions that enable AI-enabled applications to 'see' the world and make sense of it, processing images from camera feeds or digital files.

Image Classification

Training a machine learning model to classify images based on their contents.

Object Detection

Classifying individual objects within an image and identifying their location with a bounding box.

Semantic Segmentation

Classifying individual pixels in an image according to the object they belong to.

Optical Character Recognition (OCR)

Detecting and reading text in images.

Custom Vision Service

Used for training custom image classification and objective detection models.

Convolutional Neural Networks (CNN)

Filter layers extract feature maps from each image. The feature maps are flattened, and the feature values are fed into a fully connected neural network. The output layer produces a probability value for each possible class label.

Image Analysis

Image tagging, captions, model customization, and more.

Face Service

Detects faces, age, emotions, facial recognition, similarity matching, identity verification.

Natural Language Processing (NLP)

Text analysis and entity recognition, sentiment analysis speech recognition and synthesis, machine translation, semantic language modeling.

Azure AI Language

Applies AI to analyze text, understand intent, and more.

Text Analytics

Predominant Language: English. Sentiment: 88% (positive). Key Phrases, Entities.

Language Detection

Identify the language in which text is written.

Sentiment Analysis

Evaluate text and return sentiment scores and labels for each sentence.

Key Phrase Extraction

Identifying the main points around the context or context of the document(s).

Speech Recognition

The ability to detect and interpret spoken input.

Speech Synthesis

The ability to generate spoken output.

Machine Translation

Automated translation to solve the problem where organizations and individuals need to collaborate with people in other cultures and geographic locations.

Conversational AI

A solution that enables a dialog between an AI agent and a human.

Utterances

An example of something a user might say, which your application must interpret.

Entities

An item to which an utterance refers.

Intents

Represents the purpose, or goal, expressed in a user's utterance.

Document Intelligence Services

Document analysis that returns structured data representations, regions of interest and relationships. and prebuilt models.

Document Intelligence Studio

Explore the functionality using samples and your own documents.

Document Intelligence

Extract information from scanned forms in image or PDF format.

Azure AI Search

Azure’s AI-powered knowledge mining platform that finds insights at scale.

Generative AI

Create a logo for a florist business. Write Python code to add two numbers

Tokenization

The first step in training a transformer model is to decompose the training text into tokens.

LLMs

Large Language Models. Trained with large volumes of general text data. Large size can impact performance and portability

SLMs

Small Language Models. Trained with focused text data. Fewer parameters. Focused language generation capabilities in specialized contexts.

Copilots

Generative AI-powered assistants integrated into applications, often as chat interfaces.

Copilot Studio

Low-code development of copilots and plug-ins using Azure OpenAI models for generative AI.

Prompt shields

Scans for the risk of user input attacks on language models.