Home > marl > agent > agent_initlearn.m

agent_initlearn

PURPOSE ^

Initializes the agent's learning behaviour

SYNOPSIS ^

function a = agent_initlearn(a, info)

DESCRIPTION ^

 Initializes the agent's learning behaviour
  A = AGENT_INITLEARN(A, INFO)
  Initializes the agent's learning behaviour. This function must be called
  prior to any calls to agent_control.

  Parameters:
   A           - the agent
   INFO        - a structure containing other parameters that
               the agent might use during learning initialization.

  The INFO structure must contain at least the following fields:
  'episodic'       - a flag signaling whether the task is episodic
  'nagents'        - the number of agents in the task
  'statespacesize' - the size of the world state space
  'actionspacesize'- the size of the joint action space
  'imaps'          - a column vector specifying where in the world state
  vector the agent finds its own state and the common state of the world
  (i.e. state elements independent of other agents), and where on the
  joint action and reward vectors the agent finds its own action and
  reward.

  For the structure of these fields, see world_info.

  This function calls, in sequence:
  1. the actual learning initialization function of the agent.
  2. the 'init' mode of the action function, supplying the INFO structure
  on the parameters.
  3. the 'init' mode of the exploration function.

  These functions/modes will have access to the 'indices' and 'learnparam'
  fields of the agent.

  Usually, this function needs not be called directly, but is handled by
  the learning control mechanism.


  See also agent, agent_control

CROSS-REFERENCE INFORMATION ^

This function calls: This function is called by:
Generated on Wed 04-Aug-2010 16:55:08 by m2html © 2005