Lecture 2 Flashcards
What is an agent in the context of AI?
An entity with sensors to perceive the environment and actuators to manipulate it.
What are the three levels of abstraction for studying agents?
Functional (input-output), program (implementation), and architecture (system design).
What is a simple reflex agent?
An agent that selects actions based only on the current percept, using condition-action rules.
What is a model-based reflex agent?
An agent that maintains an internal state to deal with partially observable environments.
What is a goal-based agent?
An agent that reasons about the future to achieve a specific goal.
What is the PEAS acronym used for?
To define an agent’s task environment: Performance, Environment, Actuators, and Sensors.
What is the difference between a fully observable and a partially observable environment?
In a fully observable environment, the agent’s sensors can detect all relevant aspects for decision-making; in a partially observable one, they cannot.
What is a deterministic environment?
An environment where the next state is completely determined by the current state and the agent’s action.
What is a stochastic environment?
An environment where the next state involves some element of chance or unpredictability.
What is a static environment?
An environment that does not change while the agent is deciding on its next action.
What is a dynamic environment?
An environment that can change while the agent is deciding on its next action.
What is the difference between discrete and continuous environments?
Discrete environments have a finite set of states, while continuous environments have an infinite range of possible states.
What is a single-agent environment?
An environment where only one agent is taking actions.
What is a multi-agent environment?
An environment where multiple agents are acting, often with competing goals.
What is state-space representation?
A graph-based abstraction where nodes represent states and edges represent actions.
Why is state-space representation useful?
It simplifies real-world problems into manageable abstractions for decision-making.
What is a fringe in search algorithms?
A data structure that stores all the nodes that are yet to be expanded in the search tree.
How does Breadth-First Search (BFS) expand nodes?
BFS expands the shallowest unexpanded nodes level by level, using a FIFO queue.
How does Depth-First Search (DFS) expand nodes?
DFS expands the deepest unexpanded nodes first, using a LIFO stack.
What are the four performance metrics for evaluating search algorithms?
Completeness, Optimality, Time Complexity, and Space Complexity.
What is iterative deepening search?
A search method that combines BFS and DFS, expanding nodes iteratively at increasing depth levels.
Why is BFS considered complete and optimal?
Because it systematically explores all nodes at a level before moving deeper, ensuring the shortest path is found.
Why is DFS considered incomplete?
Because it can get stuck in loops and may fail to explore all nodes.
What is depth-limited search?
A variation of DFS that imposes a predefined depth limit to avoid infinite loops.
What is the horizon effect in depth-limited search?
Important outcomes beyond the depth limit may be missed, leading to suboptimal decisions.
What is the main advantage of iterative deepening search over BFS?
It uses significantly less memory while still guaranteeing optimal solutions.
What is a data-driven approach in search?
An approach where the agent starts from the current state and reasons backward to the goal.
What is a goal-driven approach in search?
An approach where the agent starts from the goal and reasons forward to the current state.
What factors determine whether to use a data-driven or goal-driven approach?
The clarity of the goal and the branching factor in each direction.
What are the main limitations of uninformed search methods?
They do not use problem-specific knowledge and have no preference for one state over another.
