net.sourceforge.jabm.learning
Class EpsilonGreedyActionSelector

java.lang.Object
  net.sourceforge.jabm.learning.EpsilonGreedyActionSelector

All Implemented Interfaces:

ActionSelector

public class EpsilonGreedyActionSelector
extends java.lang.Object
implements ActionSelector

An implementation of the epsilon-greedy action selection policy.

See:
Sutton, R. S., Barto, A. G., 1998. Reinforcement Learning: An Introduction. MIT Press.

Field Summary

static double	DEFAULT_EPSILON
protected double	epsilon The parameter representing the probability of choosing a random action on any given iteration.
protected cern.jet.random.engine.RandomEngine	prng |The pseudo-random number generator used to randomly select whether to explore and to randomly select an action when the algorithm is exploring.

Constructor Summary

EpsilonGreedyActionSelector()

EpsilonGreedyActionSelector(double epsilon, cern.jet.random.engine.RandomEngine prng)

EpsilonGreedyActionSelector(cern.jet.random.engine.RandomEngine prng)

Method Summary

int	act(int state, MDPLearner qLearner) Choose an action according to the current state and the current value estimates for each action.
double	getEpsilon()
cern.jet.random.engine.RandomEngine	getPrng()
void	setEpsilon(double epsilon)
void	setPrng(cern.jet.random.engine.RandomEngine prng)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

epsilon

protected double epsilon

The parameter representing the probability of choosing a random action on any given iteration.

prng

protected cern.jet.random.engine.RandomEngine prng

|The pseudo-random number generator used to randomly select whether to explore and to randomly select an action when the algorithm is exploring.

DEFAULT_EPSILON

public static final double DEFAULT_EPSILON

See Also:

Constant Field Values

Constructor Detail

EpsilonGreedyActionSelector

public EpsilonGreedyActionSelector(double epsilon,
                                   cern.jet.random.engine.RandomEngine prng)

EpsilonGreedyActionSelector

public EpsilonGreedyActionSelector(cern.jet.random.engine.RandomEngine prng)

EpsilonGreedyActionSelector

public EpsilonGreedyActionSelector()

Method Detail

act

public int act(int state,
               MDPLearner qLearner)

Description copied from interface: ActionSelector

Choose an action according to the current state and the current value estimates for each action.

Specified by:

act in interface ActionSelector

Parameters:

state - The current state of the MDP.

qLearner - The algorithm used to update the value estimates.

Returns:

An integer representing the action chosen (indexed from 0).

getEpsilon

public double getEpsilon()

setEpsilon

public void setEpsilon(double epsilon)

getPrng

public cern.jet.random.engine.RandomEngine getPrng()

setPrng

public void setPrng(cern.jet.random.engine.RandomEngine prng)