Artificial Intelligence Concepts, Agents, and Search Techniques

These vocabulary flashcards cover Artificial Intelligence essentials including agent types, environmental characteristics, specific AI applications, and search algorithms (uninformed and informed) including their complexities and common optimization problems.

Artificial Intelligence Main Goal

To develop systems that perform tasks requiring human intelligence, such as decision-making, reasoning, and problem solving.

Turing Test

An evaluation method commonly used to determine whether a machine can exhibit human-like intelligence, such as a chatbot convincing users it is human during a conversation.

Speech Recognition

AI technology that enables a computer to convert human speech into written text for processing.

Recommendation System

An AI application that suggests products or content to users based on their previous purchases, browsing history, and preferences.

Fraud Detection

An AI application used by banks to monitor credit card transactions and instantly detect suspicious activities to prevent financial loss.

Intelligent Agent

A system that perceives its environment through sensors and takes actions through actuators to achieve specific goals.

Percept

The complete set of inputs an agent's sensors receive at any given instant, such as camera images, GPS location, and LiDAR readings.

Partially Observable Environment

An environment where the agent cannot perceive the complete state at all times, such as a drone flying with incomplete weather information or a robot in a mall with hidden obstacles.

Episodic Environment

An environment where the agent's current action does not affect future decisions, such as a warehouse robot performing the same pick-and-place operation repeatedly.

Autonomy

The characteristic of an intelligent agent to act independently, such as a cleaning robot moving to a charging station when it detects its battery is low.

Rationality

The property of an agent selecting the action that maximizes the expected success based on the information available.

Simple Reflex Agent

An agent that selects actions based only on the current percept, ignoring history, using predefined condition-action rules.

Model-Based Reflex Agent

An agent that maintains an internal state to track the environment and predicts future positions or states before deciding its next move.

Goal-Based Agent

An agent that selects actions to reach a specific destination or objective, such as a chess-playing program evaluating move sequences to achieve checkmate.

Utility-Based Agent

An agent that chooses actions to maximize overall satisfaction or benefit, considering factors like travel time, fuel consumption, and cost.

Learning Agent

An agent that improves its performance over time by analyzing feedback and historical data from its environment.

Critic (Learning Agent)

The component of a Learning Agent that provides feedback by comparing the agent's performance against fixed, predefined standards.

Search Space

The complete set of all possible states reachable from the initial state in an AI problem.

PEAS Framework

A framework used by engineers to specify an agent's task environment, standing for Performance, Environment, Actuators, and Sensors.

Breadth-First Search (BFS)

An uninformed search logic that explores nodes level by level using a Queue; it guarantees finding the shortest path in unweighted graphs.

Depth-First Search (DFS)

An uninformed search technique that explores one branch completely using a Stack before backtracking to alternative paths.

Uniform Cost Search (UCS)

An algorithm that guarantees the optimal solution for varying positive edge costs by always expanding the node with the least cumulative path cost.

Bidirectional Search

A search strategy that simultaneously explores from both the source and the goal until the two frontiers meet.

Worst-case Time Complexity of BFS

$O(b^d)$, where $b$ is the branching factor and $d$ is the solution depth.

Worst-case Space Complexity of DFS

$O(b \times m)$, where $b$ is the branching factor and $m$ is the maximum search depth.

A* Evaluation Function

$f(n) = g(n) + h(n)$, where $g(n)$ is the actual path cost and $h(n)$ is the estimated cost (heuristic) to reach the goal.

Heuristic Value of a Goal Node

$$0$$

Local Maximum

A problem in Hill Climbing where an algorithm stops because all neighbors are lower, even though a higher global optimum exists elsewhere.

Plateau

A flat region in the search space where all neighboring states have equal heuristic values, making it difficult for an algorithm to find a direction to move.

Ridge

A narrow path in a search space where improvement is only possible through unconventional sequences of sideways or diagonal moves.

Generate and Test

A search strategy that creates possible solutions and verifies them against specific rules, such as a crossword puzzle solver checking letter combinations.

AND Node (AO* Graph)

A node type where all outgoing branches must be completed successfully to achieve the goal, such as a product passing both dimensional and electrical inspections.