Forums before death by AOL, social media and spammers... "We can't have nice things"
|    comp.ai    |    Awaiting the gospel from Sarah Connor    |    1,954 messages    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
|    Message 1,516 of 1,954    |
|    David Kinny to Randolph M. Jones    |
|    Re: Is there a specific term for agents     |
|    12 Sep 07 00:59:34    |
      From: dnk@OMIT.csse.unimelb.edu.au              "Randolph M. Jones" writes:              > Jeff Barnett wrote:       > > Behi wrote:       > >> Is there a special term in AI for agents that have memory and those       > >> that have not?       > >>       > >> -Behrang       > >>       > >       > > The terminology often used is "stateful" for those that have and       > > "stateless" for those that don't. This answer presumes that you mean       > > memory that survives from one agent interaction to its next one.              > I have also sometimes referred to agents with no internal state or       > memory as "purely reactive" agents.              I've used this term myself, but I'd suggest it's not quite the same as       the stateless/stateful distinction. It all depends on what kind of       other reactive agents exist in your preferred mental decomposition,       and what you consider to be the distinguishing feature of reactivity.              Stateless certainly implies pure reactivity. Without state there can       be no internal world model, no mental state, no dynamic goals, ...       But an agent with state can still justifiably be considered reactive.       A (deterministic) reactive agent is abstractly just a function mapping       percepts to actions. What then would you call an agent that keeps a       history of the last K percepts and maps from that history to actions?       To me this is still a reactive agent, but whether it's pure reactivity       depends on what you mean by pure.              The distinction between stateful and stateless is quite clearcut,       but once you have memory there are an infinity of things you can use       it for, and it becomes much harder to make precise distinctions,       except between obvious cases such as finite vs infinite memory.       The line drawn between reactive and deliberative agents is fuzzy,       and arbitrary even if useful, while subdivisions within these       categories tend to be more so. Then there are hybrid and layered       agent architectures to confuse the picture further.              David              [ comp.ai is moderated ... your article may take a while to appear. ]              --- SoupGate-Win32 v1.05        * Origin: you cannot sedate... all the things you hate (1:229/2)    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
(c) 1994, bbs@darkrealms.ca