[This is preliminary documentation and is subject to change.]

The QBPNetwork type exposes the following members.

Constructors

  Name Description
Protected method QBPNetwork
The base constructor for a Q-learning backpropagation network

Methods

  Name Description
Public method CheckExtraction
Indicates whether or not a rule should be extracted from the Q-learning backpropagation network
(Overrides ReinforcementTrainableBPNetwork..::..CheckExtraction()()()().)
Public method Commit
Commits the backpropagation network
(Inherited from BPNetwork.)
Protected method ComputeHiddenActivation
Computes the hidden layer activation
(Inherited from BPNetwork.)
Protected method ComputeOutputActivation
Computes the output layer activation
(Inherited from BPNetwork.)
Public method IsEligible
Indicates whether or not the component is eligible to be used
(Inherited from ClarionComponent.)
Public method Learn
Updates the network using Q-learning with standard backpropagation
(Overrides BPNetwork..::..Learn()()()().)
Protected method ModifyHiddenToOutput
Modifies the weights of the hidden to output layer
(Inherited from BPNetwork.)
Protected method ModifyInputToHidden
Modifies the weights of the input to hidden layer
(Inherited from BPNetwork.)
Public method PropagateActivation
Calculates the output activations based on the current input
(Inherited from NeuralNetwork.)
Public method ResetStatistics
Resets the match statistics
(Inherited from ReinforcementTrainableBPNetwork.)
Public method Retract (Inherited from BPNetwork.)
Public method UpdateMatchStatistics
Updates the match statistics for the reinforcement-trainable backpropagation network
(Inherited from ReinforcementTrainableBPNetwork.)

Properties

  Name Description
Public property ChosenOutput
Gets or sets the chosen output (to be used for learning)
(Inherited from ReinforcementTrainableBPNetwork.)
Protected property CommitLock
Gets the thread locker that should be used during committing process of a component
(Inherited from ClarionComponent.)
Public property DesiredOutput
Sets the desired output for the backpropagation network
(Inherited from BPNetwork.)
Protected property EligibilityCheckerDelegate
Gets the eligibility checker delegate (if one was specified during initialization, otherwise this will return null)
(Inherited from ClarionComponent.)
Protected property ExtractionMethod
Gets the extraction checker delegate (if one was specified during initialization, otherwise this will return null)
(Inherited from ReinforcementTrainableBPNetwork.)
Public property Static member Factory
Gets the factory used for generating QBPNetwork components
Public property FactoryParameters
Gets or sets the parameters that were used by the factory in order to generate the component
(Inherited from ClarionComponent.)
Public property Feedback
Gets or sets the feedback relating to the chosen output (to be used for learning)
(Inherited from ReinforcementTrainableBPNetwork.)
Public property Static member GlobalParameters
Gets the global parameters for the QBPNetwork class
Protected property Hidden
Gets the activations for the hidden layer
(Inherited from NeuralNetwork.)
Public property HiddenCount
Used to specify the number of nodes for the hidden layer
(Inherited from NeuralNetwork.)
Protected property HiddenThresholds
Gets the thresholds for the hidden layer
(Inherited from NeuralNetwork.)
Protected property HiddenToOutputWeights
Gets the hidden to output layer weight matrix
(Inherited from NeuralNetwork.)
Public property InitializeOnCommit
Specifies whether initialization operations should be performed when the component in committed
(Inherited from ClarionComponent.)
Public property Input
Gets or sets the input layer of the implicit component
(Inherited from ImplicitComponent.)
Protected property InputToHiddenWeights
Gets the input to hidden layer weight matrix
(Inherited from NeuralNetwork.)
Public property IsReadOnly
Specifies whether the component is in a read-only (i.e., immutable) state
(Inherited from ClarionComponent.)
Public property LastEligibleTimeStamp
Gets or sets the time stamp for the last time the component was eligible
(Inherited from ClarionComponent.)
Protected property MatchMethod
Gets the positive match determiner delegate (if one was specified during initialization, otherwise this will return null)
(Inherited from ReinforcementTrainableBPNetwork.)
Public property MaxQ
Gets the MaxQ value based on the new input
Public property NegativeMatch
Gets or sets the positive match statistic
(Inherited from ReinforcementTrainableBPNetwork.)
Public property NewInput
Sets the next sensory information input (used to perform Q-learning)
Public property Output
Gets the output layer of the implicit component
(Inherited from ImplicitComponent.)
Protected property OutputThresholds
Gets the thresholds for the output layer
(Inherited from NeuralNetwork.)
Public property Parameters
Gets the local parameters for this Q-learning backpropagation network
Public property PositiveMatch
Gets or sets the positive match statistic
(Inherited from ReinforcementTrainableBPNetwork.)

Explicit Interface Implementations

See Also