GPNarModel

final def !=(arg0: Any): Boolean

final def ##(): Int

final def ==(arg0: Any): Boolean

def _blockSize: Int

def _current_state: Map[String, Double]

def _errorSigma: Int

def _hyper_parameters: List[String]

def _validationSet: Seq[(DenseVector[Double], Double)]

Accessor method for validationSet

final def asInstanceOf[T0]: T0

var blockSize: Int

def blockSize_(b: Int): Unit

var caching: Boolean

def calculateEnergyPipe(h: Map[String, Double], options: Map[String, String]): DataPipe2[Seq[DenseVector[Double]], PartitionedVector, Double]

Returns a DataPipe2 which calculates the energy of data: T.

def calculateGradEnergyPipe(h: Map[String, Double]): DataPipe2[Seq[DenseVector[Double]], PartitionedVector, Map[String, Double]]

Returns a DataPipe which calculates the gradient of the energy, E(.) of data: T with respect to the model hyper-parameters.

def clone(): AnyRef

val covariance: LocalScalarKernel[DenseVector[Double]]

Underlying covariance function of the Gaussian Processes.

var current_state: Map[String, Double]

A Map which stores the current state of the system.

def data: Seq[(DenseVector[Double], Double)]

def dataAsIndexSeq(data: Seq[(DenseVector[Double], Double)]): Seq[DenseVector[Double]]

Convert from the underlying data structure to Seq[I] where I is the index set of the GP

def dataAsSeq(data: Seq[(DenseVector[Double], Double)]): Seq[(DenseVector[Double], Double)]

Convert from the underlying data structure to Seq[(I, Y)] where I is the index set of the GP and Y is the value/label type.

def energy(h: Map[String, Double], options: Map[String, String]): Double

Calculates the energy of the configuration, required for global optimization routines.

final def eq(arg0: AnyRef): Boolean

def equals(arg0: Any): Boolean

def errorSigma_(s: Int): Unit

def finalize(): Unit

val g: Seq[(DenseVector[Double], Double)]

The training data

final def getClass(): Class[_]

def getCrossKernelMatrix[U <: Seq[DenseVector[Double]]](test: U): PartitionedMatrix

def getTestKernelMatrix[U <: Seq[DenseVector[Double]]](test: U): PartitionedPSDMatrix

def getTrainKernelMatrix[U <: Seq[DenseVector[Double]]]: PartitionedPSDMatrix

def gradEnergy(h: Map[String, Double]): Map[String, Double]

Calculates the gradient energy of the configuration and subtracts this from the current value of h to yield a new hyper-parameter configuration.

def hashCode(): Int

var hyper_parameters: List[String]

Stores the names of the hyper-parameters

final def isInstanceOf[T0]: Boolean

var kernelMatrixCache: DenseMatrix[Double]

val mean: DataPipe[DenseVector[Double], Double]

The GP is taken to be zero mean, or centered.

val modelOrder: Int

def modelPredictedOutput(n: Int)(input: DenseVector[Double]): Seq[(Double, Double, Double)]

final def ne(arg0: AnyRef): Boolean

val noiseModel: LocalScalarKernel[DenseVector[Double]]

final def notify(): Unit

final def notifyAll(): Unit

val npoints: Int

var partitionedKernelMatrixCache: PartitionedPSDMatrix

def persist(state: Map[String, Double]): Unit

Cache the training kernel and noise matrices for fast access in future predictions.

def predict(point: DenseVector[Double]): Double

Predict the value of the target variable given a point.

def predictionWithErrorBars[U <: Seq[DenseVector[Double]]](testData: U, sigma: Int): Seq[(DenseVector[Double], Double, Double, Double)]

Draw three predictions from the posterior predictive distribution

def predictiveDistribution[U <: Seq[DenseVector[Double]]](test: U): MultGaussianPRV

Calculates posterior predictive distribution for a particular set of test data points.

def setState(s: Map[String, Double]): GPNarModel.this.type

Set the model "state" which contains values of its hyper-parameters with respect to the covariance and noise kernels.

final def synchronized[T0](arg0: ⇒ T0): T0

def test(testData: Seq[(DenseVector[Double], Double)]): Seq[(DenseVector[Double], Double, Double, Double, Double)]

Returns a prediction with error bars for a test set of indexes and labels.

def toString(): String

lazy val trainingData: Seq[DenseVector[Double]]

lazy val trainingDataLabels: PartitionedVector

def unpersist(): Unit

Forget the cached kernel & noise matrices.

lazy val validationDataFeatures: Seq[DenseVector[Double]]

lazy val validationDataLabels: PartitionedVector

var validationSet: Seq[(DenseVector[Double], Double)]

Setting a validation set is optional in case one wants to calculate joint marginal likelihood of the training and validation data as the objective function for hyper-parameter optimization.

def validationSet_(v: Seq[(DenseVector[Double], Double)], append: Boolean = false): Unit

Set the validation data, optionally append it to the existing validation data

final def wait(): Unit

final def wait(arg0: Long, arg1: Int): Unit

final def wait(arg0: Long): Unit

var processTargets: DataPipe[Stream[(Double, Double)], Stream[(Double, Double)]]

Assigning a value to the processTargets data pipe can be useful in cases where we need to perform operations such as de-normalizing the predicted and actual targets to their original scales.

var scoresToEnergy: DataPipe[Stream[(Double, Double)], Double]

If one uses a non empty validation set, then the user can set a custom function of the validation predictions and targets as the objective function for the hyper-parameter optimization routine.