USAII Study Book 1 - AI Essentials for Business Leaders | Fill-in-the-Blank Flashcards

Fill-in-the-blank flashcards covering key concepts from the provided notes.

The CAIC™ stands for __.

Certified Artificial Intelligence Consultant (CAIC)

The Turing test involves a human interrogating two respondents via a text interface; one is a human and the other is a __.

machine

The Total Turing Test deals with vision and __.

movement

The four major learning categories in ML are __.

Supervised learning; Unsupervised learning; Semi-supervised learning; Reinforcement learning

DL is a subset of ML based on neural networks with multiple layers; this approach is commonly called __.

deep learning

A foundation model is a general-purpose model trained on massive internet data and can be adapted for many tasks; this concept is often referred to as a __.

Foundation model

LLMs are often described as foundation models; the alternative term used for this class of models is __ models.

Foundation model

The six pillars of the GenAI Well Architected Framework are: Operational Excellence, Security & Privacy, Compliance, Reliability, Cost Optimization, and __.

System Architecture Excellence

Deployment strategies discussed include Shadow deployment, __ deployment, and Canary deployment.

A/B testing

The optimization method used to fit linear regression by minimizing squared errors is the __ method.

least squares

The General Problem Solver (GPS) was implemented using a language called __.

Information Processing Language (IPL)

In AI, two broad model types are Analytical models and __ models.

learned

The five tribes of machine learning per Pedro Domingos are Symbolists, Connectionists, Evolutionaries, Bayesians, and __.

Analogizers

K-means clustering requires choosing the number of clusters K; the method used to help select K is called the __ method.

elbow

Naive Bayes is a simple probabilistic classifier that assumes each feature is __ (independent) of every other feature.

independent

A decision tree grows by splitting data based on feature values; the criterion often used is information gain or the __ index (Gini).

Gini

Random Forest is an ensemble method that builds multiple decision trees on random samples and aggregates their predictions; this is a form of __ learning.

ensemble

A neural network architecture that processes image data with layers such as convolutional and pooling layers is called a __ neural network.

Convolutional (CNN)

Principal Component Analysis (PCA) is used to reduce dimensionality by transforming data onto a set of orthogonal axes called __.

principal components

Gradient boosting, including XGBoost, is a boosting ensemble method that sequentially adds trees to improve performance; this family is often referred to as __ boosting.

gradient

Reinforcement learning updates its policy based on __ and rewards to reinforce good behavior.

weights and rewards

The deployment strategy that gradually exposes new model updates to a subset of users is called __ deployment.

Canary

The four major components of continuous AI pipeline maintenance are CI (Continuous Integration), CD (Continuous Delivery), CT (Continuous Training), and CM ( __ ).

Continuous Monitoring

ETL stands for __, a data pipeline process used to centralize and transform data.

Extract, Transform, Load

One open-source tool for content moderation in Generative AI is __ Guardrails as provided by NVIDIA.

NeMo Guardrails

Zero-shot, one-shot, and few-shot learning are forms of __ learning.

shot learning

A 'CRISP-DM' modeling approach includes Business Understanding, Data Understanding, Data Preparation, Modeling, Evaluation, and __.

Deployment

In data storage, a 'data lake' stores raw, unstructured data; a 'data warehouse' stores structured data; a 'data lakehouse' combines features of both; the term for this integrated approach is __.

data lakehouse

The 'CI/CD' pipeline in ML is part of __, the discipline of automating the deployment and operation of machine learning systems.

MLOps

The NLP task that identifies named entities such as people, organizations, and locations is called __ (NER).

Named Entity Recognition

The diffusion model is a type of generative model used to generate high-quality data by reversing a diffusion process; this type of model is called a __ model.

diffusion

In product segmentation, Conjoint analysis helps estimate market share, sales, and __ for new designs.

profitability

ARIMA stands for AutoRegressive Integrated Moving Average; The 'I' stands for __ differencing.

Integrated