BatchRLTrainer Class Reference

#include <BatchRLTrainer.h>

Inheritance diagram for BatchRLTrainer:

Collaboration diagram for BatchRLTrainer:

Public Member Functions
	BatchRLTrainer (QFunction *qFunction, int maxExamples, float gamma, int maxTrainingIterationsPerEpisode=-1)
virtual	~BatchRLTrainer ()
virtual void	init ()
virtual void	_doTrainEpisode (DataSet *data)
	This method should be overriden by subclasses to train a single episode over the dataset. More...
Public Member Functions inherited from DataSetTrainer
	DataSetTrainer (Function *function)
	Constructor. More...
virtual	~DataSetTrainer ()
virtual void	train (DataSet *data, int maxEpisodes=10)
virtual void	trainEpisode (DataSet *data)
	Trains one episode over the dataset. More...
virtual bool	stop ()
Public Member Functions inherited from Trainer
	Trainer (Function *function)
	Constructor. More...
virtual	~Trainer ()
int	nEpisodes () const

Public Attributes
real *	targets
int	maxExamples
float	gamma
int	maxTrainingIterationsPerEpisode
RLObservation	_lastObservation
Action	_lastOrNextAction
RLObservation	_observation

Detailed Description

This class implements the Batch reinforcement learning algorithm. It trains a QFunction on a DataSet containing (s, a, r, s') tuples.

Reference: S. Lange, T. Gabel and M. Riedmiller. Batch Reinforcement Learning. In Wiering, M. and van Otterlo, M., editor, Reinforcement Learning: State of the Art, Springer, in press, 2011

See Also

TupleDataSet

Constructor & Destructor Documentation

BatchRLTrainer::BatchRLTrainer	(	QFunction *	qFunction,
		int	maxExamples,
		float	gamma,
		int	maxTrainingIterationsPerEpisode = -1
	)

BatchRLTrainer::~BatchRLTrainer ( )

virtual

Member Function Documentation

void BatchRLTrainer::_doTrainEpisode ( DataSet *  data )

virtual

This method should be overriden by subclasses to train a single episode over the dataset.

Implements DataSetTrainer.

void BatchRLTrainer::init ( )

virtual

Reimplemented from Trainer.

Member Data Documentation

RLObservation BatchRLTrainer::_lastObservation

Action BatchRLTrainer::_lastOrNextAction

RLObservation BatchRLTrainer::_observation

float BatchRLTrainer::gamma

int BatchRLTrainer::maxExamples

int BatchRLTrainer::maxTrainingIterationsPerEpisode

real* BatchRLTrainer::targets

---

The documentation for this class was generated from the following files:

• src/qualia/rl/BatchRLTrainer.h
• src/qualia/rl/BatchRLTrainer.cpp