class RecurrentNeuralNet extends Error
The RecurrentNeuralNet class feeds input in sequential time into hidden layer.
It uses parameter U, W, V in network.
where U is parameter for input x, W is for hidden layer z, and V is for output y
We have 'St = Activate (U dot x(t) + W dot x(t-1))' and
'y(t) = softmax(V dot St)'
- See also
github.com/pangolulu/rnn-from-scratch ----------------------------------------------------------------------------
- Alphabetic
- By Inheritance
- RecurrentNeuralNet
- Error
- Throwable
- Serializable
- AnyRef
- Any
- Hide All
- Show All
- Public
- Protected
Instance Constructors
- new RecurrentNeuralNet(data_dim: Int, hidden_dim: Int, bptt_truncate: Int = 4)
- data_dim
the dimension of the data space
- hidden_dim
the dimension of the hidden layer
- bptt_truncate
truncate bptt, clip to constrain the dependcy to avoid gradient vanish/explode
Value Members
- final def !=(arg0: Any): Boolean
- Definition Classes
- AnyRef → Any
- final def ##: Int
- Definition Classes
- AnyRef → Any
- final def ==(arg0: Any): Boolean
- Definition Classes
- AnyRef → Any
- final def addSuppressed(arg0: Throwable): Unit
- Definition Classes
- Throwable
- final def asInstanceOf[T0]: T0
- Definition Classes
- Any
- def bptt(x: VectoD, label: VectoD): (MatrixD, MatrixD, MatrixD)
Use back propogation through time 'bptt' to calculates dl/dV, dl/dU, dl/dW where l is the loss.
Use back propogation through time 'bptt' to calculates dl/dV, dl/dU, dl/dW where l is the loss.
- x
the input data
- label
the class labels (given ouput values)
- def calculate_loss(x: VectoD, label: VectoD): Double
Calculate the loss from the prediction of 'x' and 'label' by adding up the prediction loss among rnn layers.
Calculate the loss from the prediction of 'x' and 'label' by adding up the prediction loss among rnn layers.
- x
the input data
- label
the class labels (given ouput values)
- def calculate_total_loss(x: List[VectoD], label: List[VectoD]): Double
Calculate the total loss.
Calculate the total loss.
- x
the input data
- label
the class labels (given ouput values)
- def clone(): AnyRef
- Attributes
- protected[lang]
- Definition Classes
- AnyRef
- Annotations
- @throws(classOf[java.lang.CloneNotSupportedException]) @native() @HotSpotIntrinsicCandidate()
- final def eq(arg0: AnyRef): Boolean
- Definition Classes
- AnyRef
- def equals(arg0: AnyRef): Boolean
- Definition Classes
- AnyRef → Any
- def fillInStackTrace(): Throwable
- Definition Classes
- Throwable
- def forward_propagation(x: VectoD): List[RecurrentNeuralNetLayer]
Forward the input and generate several RNN layers.
Forward the input and generate several RNN layers.
- x
the data input
- def getCause(): Throwable
- Definition Classes
- Throwable
- final def getClass(): Class[_ <: AnyRef]
- Definition Classes
- AnyRef → Any
- Annotations
- @native() @HotSpotIntrinsicCandidate()
- def getLocalizedMessage(): String
- Definition Classes
- Throwable
- def getMessage(): String
- Definition Classes
- Throwable
- def getStackTrace(): Array[StackTraceElement]
- Definition Classes
- Throwable
- final def getSuppressed(): Array[Throwable]
- Definition Classes
- Throwable
- def hashCode(): Int
- Definition Classes
- AnyRef → Any
- Annotations
- @native() @HotSpotIntrinsicCandidate()
- def initCause(arg0: Throwable): Throwable
- Definition Classes
- Throwable
- final def isInstanceOf[T0]: Boolean
- Definition Classes
- Any
- final def ne(arg0: AnyRef): Boolean
- Definition Classes
- AnyRef
- final def notify(): Unit
- Definition Classes
- AnyRef
- Annotations
- @native() @HotSpotIntrinsicCandidate()
- final def notifyAll(): Unit
- Definition Classes
- AnyRef
- Annotations
- @native() @HotSpotIntrinsicCandidate()
- def printStackTrace(arg0: PrintWriter): Unit
- Definition Classes
- Throwable
- def printStackTrace(arg0: PrintStream): Unit
- Definition Classes
- Throwable
- def printStackTrace(): Unit
- Definition Classes
- Throwable
- var rvm: RandomMatD
- def setStackTrace(arg0: Array[StackTraceElement]): Unit
- Definition Classes
- Throwable
- def sgd_step(x: VectoD, label: VectoD, learning_rate: Double): MatrixD
Stochastic gradient descent step.
Stochastic gradient descent step.
- x
the input data
- label
the class labels (given ouput values)
- final def synchronized[T0](arg0: => T0): T0
- Definition Classes
- AnyRef
- def toString(): String
- Definition Classes
- Throwable → AnyRef → Any
- def train(x: List[VectoD], label: List[VectoD], rate: Double = 500.0, nepoch: Int, eval_loss_after: Int = 5): RecurrentNeuralNet
Train the model by iterating throught the training set by sgd and adjusting the learning rate.
Train the model by iterating throught the training set by sgd and adjusting the learning rate.
- x
the input data
- label
the class labels (given ouput values)
- rate
the initial learning rate (gradient multiplier)
- nepoch
number of epoch
- eval_loss_after
number of epoch to conduct evaluation
- val u: MatrixD
- val v: MatrixD
- val w: MatrixD
- final def wait(arg0: Long, arg1: Int): Unit
- Definition Classes
- AnyRef
- Annotations
- @throws(classOf[java.lang.InterruptedException])
- final def wait(arg0: Long): Unit
- Definition Classes
- AnyRef
- Annotations
- @throws(classOf[java.lang.InterruptedException]) @native()
- final def wait(): Unit
- Definition Classes
- AnyRef
- Annotations
- @throws(classOf[java.lang.InterruptedException])
Deprecated Value Members
- def finalize(): Unit
- Attributes
- protected[lang]
- Definition Classes
- AnyRef
- Annotations
- @throws(classOf[java.lang.Throwable]) @Deprecated
- Deprecated