Untitled Quiz

Environment Types

• Environments can be categorized by several properties, including sequential vs. episodic, static vs. dynamic, discrete vs. continuous, known vs. unknown, single-agent vs. multi-agent, and deterministic vs. stochastic.

Sequential vs. Episodic Environments

• In sequential environments, current decisions can affect all future decisions. Examples include chess and taxi driving, where short-term actions have long-term consequences.
• Episodic environments are simpler since the agent doesn't need to think ahead. Each episode is independent of the others.

Static vs. Dynamic Environments

• A dynamic environment changes while the agent is deliberating. Taxi driving is dynamic because the cars are constantly moving.
• A static environment remains unchanged while the agent is deciding on an action. Crossword puzzles are static.
• Semidynamic environments don't change themselves, but the agent's performance score does. Chess, played with a clock, fits this category.

Discrete vs. Continuous Environments

• This distinction can apply to the state of the environment, time, percepts, and actions.
• A discrete environment has a finite number of distinct states. Chess is considered discrete.
• A continuous environment has an infinite number of states. Taxi driving is a continuous-state and continuous-time problem.

Known vs. Unknown Environments

• This distinction refers to the agent's knowledge of the environment's "laws of physics."
• In a known environment, the outcomes of all actions are known.
• In an unknown environment, the agent must learn how the environment works to make good decisions.

Partially Observable Environments

• An environment might be partially observable due to noisy or inaccurate sensors, or because information is missing from the sensor data.
• For example, a vacuum cleaner with a local dirt sensor can't know if there's dirt in other squares.

Unobservable Environments

• An environment where the agent has no sensors at all is considered unobservable.

Single-Agent vs. Multi-Agent Environments

• A single-agent environment involves only one agent. Solving a crossword puzzle alone is an example.
• A multi-agent environment involves multiple agents interacting. Chess is a two-agent environment where each tries to maximize its performance.
• Competitive multi-agent environments have agents with conflicting goals. In chess, maximizing one agent's score minimizes the other's.
• Partially cooperative multi-agent environments have agents with aligned goals in some aspects. In taxi driving, avoiding collisions benefits all drivers.

Deterministic vs. Stochastic Environments

• A deterministic environment has a completely predictable next state based on the current state and the action taken.
• A stochastic environment has unpredictable outcomes, meaning the next state cannot be perfectly predicted. Most real-world situations are stochastic.

Environment Type Summary

Agent Functions and Programs

• The goal of AI is to design an agent program that implements the agent function, which maps percepts to actions.
• The program runs on an architecture, which includes physical sensors and actuators.
• Agent = Architecture + Program

Table-Driven Agent

• A table-driven agent uses a table that maps every possible percept sequence to the appropriate action.
• Drawbacks include:
  • Huge table size
  • Time-consuming table construction
  • Lack of autonomy
  • Long learning time

Simple Reflex Agents

• Simple reflex agents select actions based only on the current percept, ignoring past history.
• Condition-action rules are used to map percepts to actions.
• INTERPRET-INPUT function creates a state description from the percept.
• RULE-MATCH function finds the first rule that matches the current state description.