XPost: comp.ai.alife   
   From: christian.guttmann@csse.monash.edu.au   
      
   > >  > Is collaboration possible without communication?   
   >   
   > Hi Christian,   
   >   
   > The large mass of multi-agent (reinforcement) learning is about learning   
   > some task with multiple agents, without communicating at all.   
   > Many systems work by 'shared rewards', so that agents will learn to do   
   > their actions in the benefit of the whole system. More sophisticated   
   > approaches use for example 'wonderful life utility' in which the   
   > difference between system+agent1 and system-without-agent1 for example.   
   >   
      
   Thanks Martijn.   
      
   The notion of collaboration among agents is somewhat integrated by the   
   designer of such learning system. There may be no explicit and direct   
   communication during run time, but a model of collaborative utility is   
   captured when designing the system and before reinforcement learning   
   takes place , isn't it?   
      
   cheers, Christian   
      
   [ comp.ai is moderated.  To submit, just post and be patient, or if ]   
   [ that fails mail your article to , and ]   
   [ ask your news administrator to fix the problems with your system. ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)