rationality only concerns with the end decisions that are made, not the thought process behind them
“brains are to intelligent, as wings are to flight” This is often quoted to posit the practicality in building intelligent agents should not be “exactly” the same as a brain. As flight does not imitate the mechanics of wings but takes advantage of aerodynamics. However, I am not a proponent of this proposition. I believe the imitation of brains lead superior intelligent agents.
Environment Properties
fully-observable/accessible v partially-observable/in-accessible
accessible - agent has complete state of the environment
in-accessible -
deterministic v non-deterministic/stochastic v strategic
deterministic - next state of environment is completely determined by the current state and the agent’s action
strategic - environment is deterministic except for the actions of other agents
episodic v non-episodic/sequential
episodic - in an episodic environment, the agent’s experience is divided into “episodes.” Each episode consists of the agent perceiving and then acting. The quality of its action depends just on the episode itself, because subsequent episodes do not depend on what actions occur in previous episodes. Episodic environments are much simpler because the agent does not need to think ahead only on the episode itself
static v dynamic v semi-dynamic
static - environment is unchanging
dynamic - environment is changing
semi-dynamic - environment does not change with time but the agent’s performance score does
discrete v continuous
discrete - environment provided fixed number of distinct percepts, actions, and/or environment states
continuous - environment has a continuous number of distinct percepts, actions, and/or environment states
single-agent v multi-agent
single - agent operating alone in an environment
known v unknown
known - the agent knows the rules of the environment
4 types of Agents
simple reflex agents
uses (if condition then action)
environments need to be fully-observable
agents that keep track of the world
agents have internal states
uses (if condition then action)
environments can be partially-observable
goal-based agents
does not use (if condition then action)
but actions are chosen based on reaching towards goal
agent actions are towards a goal
utility-based agents
a goal-based agent with a metric to assist the agent in moving toward the goal faster