mlpy.mdp.discrete.LeastVisitedBonusExplorer

class mlpy.mdp.discrete.LeastVisitedBonusExplorer(rmax, func, thresh=None)

Bases: mlpy.mdp.discrete.RMaxExplorer

Least visited bonus explorer, a RMax based exploration model.

Least visited bonus exploration only goes into exploration mode whether it is predicted that only states with rewards less than a given threshold can be reached. Once in exploration mode, states that have been visited least are given a bonus of RMax to drive exploration.

Parameters:

rmax : float

The maximum achievable reward.

func : callable

Callback function to retrieve the minimum number of times a state has been visited.

thresh : float

If all states that can be reached from the current state have a value less than the threshold, exploration mode is turned on.

Attributes

mid	The module’s unique identifier.

Methods

activate([qvalues])	Turn on exploration mode.
deactivate()	Turn off exploration mode.
load(filename)	Load the state of the module from file.
save(filename)	Save the current state of the module to file.
update(model)	Update the reward model.