01 Introduksjon Flashcards
Vanlig “sykel” i kunstig intelligens
perception-action cycle
parts of an AI system
environment
agent: sensors -> actuators
når sensorene alltid kan se hele tilstanden til systemet
fully observable (environment)
fully observable (environment)
when the sensors (agent) always can see the entire state of the environment
når agenten må bruke minne for å gradvis danne seg et bilde av omgivelsene
partially observable (environment)
partially observable (environment)
when the sensors (agent) can only see a fraction of the state, yet memorizing past measurements
deterministic environment
an environment in which agents’ actions uniquely determine the outcome (no randomness)
an environment in which agents’ actions uniquely determine the outcome
deterministic environment
stochastic environment
an environment in which a certain amount of randomness is involved
an environment in which a certain amount of randomness is involved
stochastic environment
discrete environment
an environment where the space of possible actions and things you could sense is finite
an environment where the space of possible actions and things you could sense is finite
discrete environment
continuous environment
an environment where the space of possible actions or things you could sense may be infinite
an environment where the space of possible actions or things you could sense may be infinite
continuous environment
benign environment
an environment that does not try to obstruct the agent
an environment that does not try to obstruct the agent
benign environment
adversarial environment
an environment that actively tries to obstruct the agent
an environment that actively tries to obstruct the agent
adversarial environment
AI as _____ management
AI as uncertainty management
sources of uncertainty
sensor limits stochastic environments ignorance adversaries laziness ...
key attributes of AI environments
fully vs partially observable
deterministic vs stochastic
discrete vs continuous
benign vs adversarial
