@TechReport{crutchfield01a,
  author =	 {James P. Crutchfield and David P. Feldman },
  title =	 {Synchronizing to the Environment: Information
                  Theoretic Constraints on Agent Learning},
  googleid =	 {aQWcwL8nLvEJ:scholar.google.com/},
  institution =	 {Santa Fe Institute},
  year =	 2001,
  url =		 {http://arxiv.org/pdf/nlin.AO/0103038},
  abstract =	 {We show that the way in which the Shannon entropy of
                  sequences produced by an information source
                  converges to the source's entropy rate can be used
                  to monitor how an intelligent agent builds and
                  effectively uses a predictive model of its
                  environment. We introduce natural measures of the
                  environment's apparent memory and the amounts of
                  information that must be (i) extracted from
                  observations for an agent to synchronize to the
                  environment and (ii) stored by an agent for optimal
                  prediction. If structural properties are ignored,
                  the missed regularities are converted to apparent
                  randomness. Conversely, using representations that
                  assume too much memory results in false
                  predictability. },
  keywords =     {learning complexity},
  comment =	 {They use a contrived model of the environment---a
                  hidden Markov model that outputs signals to the
                  agent upon any transition. The agent's learning task
                  is to determine which state the markov model is
                  in. I cannot think of any multiagent system whose
                  learning goal is to determine which "hidden state"
                  the environment is in. These results seem more
                  useful when modeling other agents that are
                  themselves markovian processes.},
  cluster = 	 {17378871716593010025}
}