Glossary of Artificial Intelligence Terms

AI Definition

• Artificial Intelligence (AI) – The field of creating machines that simulate human intelligence, e.g., a voice assistant answering questions.

• Intelligent Behavior – The ability to learn and make decisions, e.g., Google Maps choosing the fastest route.

AI Problems

• AI Problem – A task requiring intelligent decision-making, e.g., diagnosing a disease from symptoms.

• Problem Solving – Finding steps to reach a goal, e.g., solving a maze by choosing correct turns.

Foundations of Artificial Intelligence

• Philosophy – Provides logical reasoning concepts, e.g., using logic to determine if a statement is true.

• Mathematics – Supplies formal tools like probability, e.g., calculating the likelihood of rain.

• Psychology – Helps model human thinking, e.g., understanding how people learn languages.

• Neuroscience – Inspires neural networks, e.g., brain-like models used in image recognition.

• Computer Science – Provides algorithms and data structures, e.g., using trees for decision making.

• Control Theory – Manages system behavior, e.g., cruise control maintaining car speed.

• Linguistics – Enables language understanding, e.g., chatbots interpreting user sentences.

AI Techniques

• AI Technique – A method to solve problems efficiently, e.g., using rules to play chess.

• Heuristic – A shortcut strategy, e.g., choosing the nearest restaurant to save time.

AI Models

• Symbolic Model – Uses symbols and rules, e.g., expert systems for medical diagnosis.

• Subsymbolic Model – Uses numerical learning, e.g., neural networks recognizing faces.

Defining Problem as a State Space Search

• State – A specific situation in a problem, e.g., current position in a GPS route.

• State Space – All possible situations, e.g., all routes between two cities.

• Initial State – The starting situation, e.g., starting location in navigation.

• Goal State – The desired outcome, e.g., reaching the destination.

• Operator – An action that changes state, e.g., turning left or right while driving.

Production System

• Production Rule – A condition-action rule, e.g., IF traffic is heavy THEN take alternate route.

• Production System – A rule-based problem solver, e.g., automated customer support systems.

• Working Memory – Stores current facts, e.g., active orders in a food delivery app.

• Control Strategy – Chooses which rule to apply, e.g., prioritizing urgent alerts first.

Intelligent Agents: Agents and Environments

• Agent – An entity that acts intelligently, e.g., a self-driving car.

• Environment – Everything the agent interacts with, e.g., roads, traffic, and pedestrians.

• Percept – Information received by an agent, e.g., camera detecting a stop sign.

• Action – A step taken by an agent, e.g., applying brakes.

Characteristics of Intelligent Agents

• Rationality – Choosing the best action, e.g., a robot avoiding obstacles efficiently.

• Autonomy – Operating independently, e.g., a vacuum robot cleaning on its own.

• Reactivity – Responding to changes, e.g., adjusting speed when traffic increases.

• Proactiveness – Taking initiative, e.g., scheduling reminders before deadlines.

Search Methods

• Search Algorithm – A method to explore solutions, e.g., finding a path in a map.

• Uninformed Search – Search without guidance, e.g., checking every door to find an exit.

• Informed Search – Search using heuristics, e.g., following signboards to find an airport.

• Breadth-First Search – Exploring all nearby options first, e.g., checking closest shops first.

• Depth-First Search – Exploring one path fully, e.g., following one road until it ends.

Issues in the Design of Search Problems

• Search Space Size – Number of possible choices, e.g., all chess board positions.

• Branching Factor – Choices available at each step, e.g., possible moves in a game.

• Optimality – Finding the best solution, e.g., shortest travel route.

• Completeness – Guaranteeing a solution, e.g., a maze solver that always finds an exit.

• Time Complexity – Time taken to solve, e.g., fast vs slow route planning.

• Space Complexity – Memory required, e.g., storing all visited locations in navigation.

Knowledge Representation Issues

• Knowledge Representation – The method of encoding information so an AI system can reason about it, e.g., representing traffic rules in a driving system.

• Knowledge Representation Issues – Challenges in expressing knowledge accurately and efficiently, e.g., representing common-sense human reasoning in machines.

Mapping

• Mapping – The process of converting real-world knowledge into a formal representation, e.g., mapping a city map into a graph for navigation.

Frame Problem

• Frame Problem – The difficulty of representing what remains unchanged after an action, e.g., knowing a room’s color stays the same after moving a chair.

Predicate Logic

• Predicate Logic – A formal system using predicates and quantifiers to represent relationships, e.g., Student(John).

Facts in Logic

• Fact – A basic assertion assumed to be true, e.g., Human(Socrates).

Representing Instance and ISA Relationship

• Instance Relationship – Specifies an object belongs to a class, e.g., Tweety is a Bird.

• ISA Relationship – Represents class-subclass hierarchy, e.g., Bird ISA Animal.

Resolution

• Resolution – A rule of inference used to derive conclusions by refutation, e.g., proving a statement true by showing its negation leads to contradiction.

Procedural and Declarative Knowledge

• Procedural Knowledge – Knowledge of how to perform tasks, e.g., steps to cook a recipe.

• Declarative Knowledge – Knowledge of facts and statements, e.g., knowing water boils at 100°C.

Matching

• Matching – The process of comparing patterns with stored knowledge, e.g., matching symptoms to disease rules.

Control Knowledge

• Control Knowledge – Knowledge that guides the reasoning process, e.g., deciding which medical test to perform first.

Symbolic Reasoning Under Uncertainty

• Symbolic Reasoning Under Uncertainty – Reasoning with incomplete or uncertain symbolic information, e.g., diagnosing illness with missing symptoms.

Non-Monotonic Reasoning

• Non-Monotonic Reasoning – Reasoning where conclusions can be withdrawn when new information appears, e.g., assuming a bird flies until learning it is a penguin.

Statistical Reasoning

• Statistical Reasoning – Reasoning using probabilities and data, e.g., predicting weather based on historical data.

Game Playing

• Game Playing – AI techniques used to make optimal decisions in competitive environments, e.g., a computer playing chess.

Minimax Search

• Minimax Search – A decision strategy that minimizes the possible loss assuming an optimal opponent, e.g., choosing a chess move assuming the opponent plays best.

Alpha-Beta Cut-offs

• Alpha-Beta Cut-offs – An optimization of minimax that prunes irrelevant branches, e.g., skipping moves that cannot affect the final chess outcome.

Natural Language Processing

• Natural Language Processing (NLP) – The ability of computers to understand and generate human language, e.g., chatbots answering customer queries.

Learning

5. Learning – The process by which an AI system improves performance from experience, e.g., a spam filter improving over time.

Explanation-Based Learning

• Explanation-Based Learning – Learning by generalizing from a single example using domain knowledge, e.g., learning rules from one successful diagnosis.

Discovery

• Discovery – Learning new patterns or concepts from data, e.g., discovering buying patterns in supermarket data.

Analogy

• Analogy – Learning by transferring knowledge from a similar situation, e.g., solving a new math problem using a previously solved one.

Neural Net Learning

• Neural Network Learning – Learning by adjusting weights in interconnected neurons, e.g., recognizing handwritten digits.

Genetic Learning

• Genetic Learning – Learning using evolutionary principles like selection and mutation, e.g., optimizing delivery routes through generations.

Fuzzy Logic Systems

• Fuzzy Logic System – A system that handles imprecise information using degrees of truth, e.g., automatic air conditioners adjusting temperature.

Perception and Action

• Perception and Action – The cycle of sensing the environment and responding to it, e.g., a robot detecting obstacles and moving around them.

Expert Systems

• Expert System – A computer system that emulates human expert decision-making, e.g., medical diagnosis software.

Inference in Bayesian Networks

• Bayesian Network Inference – Probabilistic reasoning using conditional dependencies, e.g., calculating disease probability from symptoms.

K-Means Clustering Algorithm

• K-Means Clustering – An unsupervised algorithm that groups similar data points, e.g., customer segmentation in marketing.

Machine Learning

• Machine Learning – A subset of AI that enables systems to learn patterns from data, e.g., movie recommendation systems.