Jump to content

Intelligent agent

From Wikipedia, the free encyclopedia
This is an old revision of this page, as edited by Mneser (talk | contribs) at 11:03, 20 September 2005. The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.

re merge: Please refer to discussion.

In computer science an Intelligent Agent (IA) refers to a software agent that exhibits some form of artificial intelligence. While the working of software agents used for operator assistance or data mining (sometimes refered to as bots), are often based on fixed pre-programmed rules, "intelligent" implies the ability to adapt and learn.

In some literature IAs are also refered to as autonomous intelligent agents, although autonomy is generally assumed. According to Nicola Kasabov [1] IA systems should exibit the following characteristics:

  • learn and improve through interaction with the environment (embodiment)
  • adapt online and in real time
  • learn quickly from large amounts of data
  • accommodate new problem solving rules incrementally
  • have memory based exemplar storage and retrieval capacities
  • have parameters to represent short and long term memory, age, forgetting, etc.
  • be able to analyse itself in terms of behaviour, error and success

Further development in this field include fuzzy agents (IA implimentations with fuzzy logic) and cognitive agents (IA considered to be self-aware).

References

[2] Bridgeport [3] Coneural