AI A Modern Approach - Chapter 2

agent

anything that perceives its environment through sensors and acts upon that environment through actuators

percept

the agent's perceptual inputs at any given instant

percept sequence

complete history of everything the agent has ever perceived

external agent function

maps any given percept sequence to an action

internal agent function

the agent's program

performance measure

notion of desirability that evaluates any given sequence of environment states

rational agent

For each possible percept sequence, a rational agent should select an action that is expected to maximize its performance measure, given the evidence provided by the percept sequence and whatever built-in knowledge the agent has.

omniscient agent

knows the actual outcome of its actions and can act accordingly; but omniscience is impossible in reality

information gathering

doing actions in order to modify future percepts

exploration

example of information gathering

autonomy

learn what it can to compensate for partial or incorrect prior knowledge

task environment

problems to which rational agents are the solutions

PEAS

Performance, Environment, Actuators, Sensors

software agent (softbot)

software robots

stochastic

the environment is not determine by just the current state and the action executed by the agent

uncertain

environment that is not fully observable or not deterministic

nondeterministic

environment in which actions are characterized by their possible outcomes

episodic task

the agent's experience is divided into atomic episodes, in each episode the agent receives a percept and then performs a single action. Crucially, the next episode does not depend on the actions taken in previous episodes

task environment categories

observable, agents, deterministic, episodic, static, discrete

environment class

environments are drawn from this that way the agent is not able to take advantage of a single environment's particular characteristics

environment generator

class that selects particular environments win which to run the agent

agent program

implements the agent function - the mapping from percepts to actions

architecture

some sort of computing device with physical sensors actuators (agent = architecture + program)

simple reflex agent

select actions on the basis of the current percept, ignoring the rest of the percept history

condition-action rule, situation-action rule, production

if car-in-front-is-braking then initiate-braking.

internal state

in model-based reflex agents the agent should maintain this sate that depends on the percept history and thereby reflects at least some of the unobserved aspects of the current state

model-based agent

an agent that tracks the world

goal-based agents

more flexible because the knowledge that supports its decisions is represented explicitly and can be modified

utility

how "happy" and "unhappy" is the state of getting to the end goal

utility function

internalization of the performance measure

expected utility

the utility the agent expects to derive, on average, given the probabilities and utilities of each outcome

learning element

responsible for making improvements

performance element

responsible for selecting external actions

critic

element that provides feedback on how the agent is doing, the learning element uses it

problem generator

part responsible for suggesting actions that will lead to new and informative experiences

factored representation

splits up each state into a fixed set of variables or attributes, each of which can have a value

structured representation

representation in which objects and their various and varying relationships can be described explicitly

expressiveness

axis along which atomic, factored, and structured representations lie is this axis increasing