Home > marl > agent > actfuns > asfq_greedyact.m

asfq_greedyact

PURPOSE ^

Implements deterministic greedy action selection for adaptive state focus Q-learning

SYNOPSIS ^

function [a, action] = asfq_greedyact(a, state, actions, rewards, params)

DESCRIPTION ^

Implements deterministic greedy action selection for adaptive state focus Q-learning
  [A, ACTION] = ASFQ_GREEDYACT(A, STATE, ACTIONS, REWARDS, PARAMS)
  Implements deterministic greedy action selection, to be combined with
  the adaptive state focus Q-learning algorithm. The action with the
  highest Q-value is chosen (breaking ties randomly). The action selection
  mechanism takes into account the learning mode of the agent
  (single-agent, or full state), and thus the structure of the Q-table.

  Supports discrete states and actions, with 1 action variable only.

  Requires the size of the Q-table to be cached under a field named
  'qsize'.


  See also agent_act, asfq

CROSS-REFERENCE INFORMATION ^

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