NeuralDenseCrfWorker

Instance Constructors

new NeuralDenseCrfWorker(lambdas: Array[Double], nls: Int, nfs: Int, segSize: Int, gPrior: Double, nNfs: Int, nGates: Int, opts: Options)

Type Members

type Matrix = Array[Array[Double]]

Definition Classes
PotentialScoring
type Tensor = Array[Matrix]

Definition Classes
PotentialScoring

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
val activations: Array[Array[Double]]

Definition Classes
NeuralDenseCrf
var adjustible: Boolean

When set to true, the Crf will allow the state-space to be dynamically sized - i.e.
When set to true, the Crf will allow the state-space to be dynamically sized - i.e. the number of states is dependent on each sequence

Definition Classes
Crf
var alpha: Matrix

Alpha values.
Alpha values. Need values for each segment length for each label (in general, Semi-CRF case)

Definition Classes
Crf
final def asInstanceOf[T0]: T0

Definition Classes
Any
def assign(v1: Array[Double], f: (Double) ⇒ Double): Unit

Attributes
protected
Definition Classes
Crf
def assign1(v1: Array[Double], v2: Array[Double], f: (Double, Double) ⇒ Double): Unit

Attributes
protected
Definition Classes
Crf
def backwardPass(iseq: Seq[AbstractInstance]): Unit

Attributes
protected
Definition Classes
Crf
var beta: Matrix

Beta values.
Beta values. Need values for each segment length for each label (in general, Semi-CRF case)

Definition Classes
Crf
def clone(): AnyRef

Attributes
protected[java.lang]
Definition Classes
AnyRef
Annotations
@throws( ... )
def computeGateActivations(lambdas: Array[Double], acts: Array[Double], wActs: Array[Double], numFs: Int, gateIdx: Int, nls: Int, nGates: Int, nNfs: Int, inst_features: Array[Array[Feature]]): Unit

Definition Classes
NeuralStochasticCrfScoring
def computeScores(inst_features: Array[Array[Feature]], takeExp: Boolean): Unit

Definition Classes
NeuralDenseCrf → Crf
def computeScores(ri: Matrix, mi: Tensor, lambdas: Array[Double], acts: Array[Double], wActs: Array[Double], numFs: Int, gateIdx: Int, nls: Int, nGates: Int, nNfs: Int, inst_features: Array[Array[Feature]], takeExp: Boolean): Unit

Definition Classes
NeuralStochasticCrfScoring
final def computeScores(ri: Matrix, mi: Tensor, inst_features: Array[Array[Feature]], takeExp: Boolean, nls: Int, lambdas: Array[Double]): Unit

Definition Classes
PotentialScoring
val curA: Array[Double]

Current alpha values used for Forward-Backward computation
Current alpha values used for Forward-Backward computation

Definition Classes
Crf
var curNls: Int

Definition Classes
Crf
final def eq(arg0: AnyRef): Boolean

Definition Classes
AnyRef
def equals(arg0: Any): Boolean

Definition Classes
AnyRef → Any
val featureExpectations: Array[Double]

Definition Classes
DenseCrf
def finalize(): Unit

Attributes
protected[java.lang]
Definition Classes
AnyRef
Annotations
@throws( classOf[java.lang.Throwable] )
def forwardPass(iseq: IndexedSeq[AbstractInstance]): Double

Attributes
protected
Definition Classes
NeuralDenseCrf → DenseCrf → Crf
val gPrior: Double

The Gaussian prior variance used as a regularizer
The Gaussian prior variance used as a regularizer

Definition Classes
Crf
val gateWeightIdx: Int

Definition Classes
NeuralDenseCrf
final def getClass(): Class[_]

Definition Classes
AnyRef → Any
def getCoreModel(): CoreModel

Definition Classes
Crf → Trainable
def getGradient(l2: Boolean, seqAccessor: AccessSeq[AbstractInstance]): Option[Double]

Definition Classes
DenseCrf
def getGradient(seqAccessor: AccessSeq[AbstractInstance]): Option[Double]

Definition Classes
DenseCrf → Crf → Trainable
def getLambdas: Array[Double]

Definition Classes
Trainable
def gradOfSeq(iseq: IndexedSeq[AbstractInstance]): Double

Definition Classes
DenseCrf
val gradient: Array[Double]

Definition Classes
DenseCrf
def hashCode(): Int

Definition Classes
AnyRef → Any
def initialize(): Unit

Definition Classes
NeuralDenseCrf → Crf → Trainable
val invSigSqr: Double

The value of the inverse square of the Gaussian prior
The value of the inverse square of the Gaussian prior

Definition Classes
Crf
final def isInstanceOf[T0]: Boolean

Definition Classes
Any
val lambdas: Array[Double]

Parameter (lambda) vector
Parameter (lambda) vector

Definition Classes
Crf → Trainable
final def matrixMult(mat: Matrix, vec: Array[Double], rvec: Array[Double], alpha: Double, beta: Double, trans: Boolean): Unit

Definition Classes
PotentialScoring
val mi: Tensor

For each segment size, the mi matrix holds transition scores for adjacent labels
For each segment size, the mi matrix holds transition scores for adjacent labels

Definition Classes
Crf
val nGates: Int

Number of neural gates per label (for NeuralCrf)
Number of neural gates per label (for NeuralCrf)

Definition Classes
Crf
val nNfs: Int

Number of neural gate input features (for NeuralCrf)
Number of neural gate input features (for NeuralCrf)

Definition Classes
Crf
final def ne(arg0: AnyRef): Boolean

Definition Classes
AnyRef
val newA: Array[Double]

Alpha values at the next position used for Forward-Backward computation
Alpha values at the next position used for Forward-Backward computation

Definition Classes
Crf
val nfs: Int

Number of features
Number of features

Definition Classes
Crf
val nls: Int

Number of labels/states
Number of labels/states

Definition Classes
Crf
final def notify(): Unit

Definition Classes
AnyRef
final def notifyAll(): Unit

Definition Classes
AnyRef
val numInputFeatures: Int

Definition Classes
NeuralDenseCrf
val numParams: Int

Definition Classes
Crf → Trainable
def regularize(): Double

Definition Classes
DenseCrf
def reset(all: Boolean, slen: Int): Unit

Attributes
protected
Definition Classes
Crf
def resetParameters(): Unit

Definition Classes
Crf
val ri: Matrix

For each segment size (general case) the ri matrix holds state scores for each label
For each segment size (general case) the ri matrix holds state scores for each label

Definition Classes
Crf
var scale: Array[Double]

An array of scaling coefficients to avoid underflow without having to do computations in log space.
An array of scaling coefficients to avoid underflow without having to do computations in log space. See Manning and Schutze Chapter 9 for details (there in the context of HMMs)

Definition Classes
Crf
val segSize: Int

The size of segments.
The size of segments. Sizes greater than 1 indicate the model is a semi-CRF

Definition Classes
Crf
final def setMatrix(m: Matrix, v: Double = 0.0): Unit

Definition Classes
PotentialScoring
final def setTensor(t: Tensor, v: Double = 0.0): Unit

Definition Classes
PotentialScoring
final def synchronized[T0](arg0: ⇒ T0): T0

Definition Classes
AnyRef
val tmp: Array[Double]

Definition Classes
Crf
def toString(): String

Definition Classes
AnyRef → Any
def train(a: AccessSeq[AbstractInstance], x: Int, modelIterFn: Option[(CoreModel, Int) ⇒ Unit] = None): Nothing

Definition Classes
DenseWorker → Crf → Trainable
def train(a: AccessSeq[AbstractInstance]): Nothing

Definition Classes
DenseWorker → Crf
def vecSum(vec: Array[Double]): Double

Attributes
protected
Definition Classes
Crf
final def wait(): Unit

Definition Classes
AnyRef
Annotations
@throws( ... )
final def wait(arg0: Long, arg1: Int): Unit

Definition Classes
AnyRef
Annotations
@throws( ... )
final def wait(arg0: Long): Unit

Definition Classes
AnyRef
Annotations
@throws( ... )
val weightedActivationPartials: Array[Double]

Definition Classes
NeuralDenseCrf

Related Doc: package crf

class NeuralDenseCrfWorker extends NeuralDenseCrf with DenseWorker

Instance Constructors

new NeuralDenseCrfWorker(lambdas: Array[Double], nls: Int, nfs: Int, segSize: Int, gPrior: Double, nNfs: Int, nGates: Int, opts: Options)

Type Members

type Matrix = Array[Array[Double]]

type Tensor = Array[Matrix]

Value Members

final def !=(arg0: Any): Boolean

final def ##(): Int

final def ==(arg0: Any): Boolean

val activations: Array[Array[Double]]

var adjustible: Boolean

var alpha: Matrix

final def asInstanceOf[T0]: T0

def assign(v1: Array[Double], f: (Double) ⇒ Double): Unit

def assign1(v1: Array[Double], v2: Array[Double], f: (Double, Double) ⇒ Double): Unit

def backwardPass(iseq: Seq[AbstractInstance]): Unit

var beta: Matrix

def clone(): AnyRef

def computeGateActivations(lambdas: Array[Double], acts: Array[Double], wActs: Array[Double], numFs: Int, gateIdx: Int, nls: Int, nGates: Int, nNfs: Int, inst_features: Array[Array[Feature]]): Unit

def computeScores(inst_features: Array[Array[Feature]], takeExp: Boolean): Unit

def computeScores(ri: Matrix, mi: Tensor, lambdas: Array[Double], acts: Array[Double], wActs: Array[Double], numFs: Int, gateIdx: Int, nls: Int, nGates: Int, nNfs: Int, inst_features: Array[Array[Feature]], takeExp: Boolean): Unit

final def computeScores(ri: Matrix, mi: Tensor, inst_features: Array[Array[Feature]], takeExp: Boolean, nls: Int, lambdas: Array[Double]): Unit

val curA: Array[Double]

var curNls: Int

final def eq(arg0: AnyRef): Boolean

def equals(arg0: Any): Boolean

val featureExpectations: Array[Double]

def finalize(): Unit

def forwardPass(iseq: IndexedSeq[AbstractInstance]): Double

val gPrior: Double

val gateWeightIdx: Int

final def getClass(): Class[_]

def getCoreModel(): CoreModel

def getGradient(l2: Boolean, seqAccessor: AccessSeq[AbstractInstance]): Option[Double]

def getGradient(seqAccessor: AccessSeq[AbstractInstance]): Option[Double]

def getLambdas: Array[Double]

def gradOfSeq(iseq: IndexedSeq[AbstractInstance]): Double

val gradient: Array[Double]

def hashCode(): Int

def initialize(): Unit

val invSigSqr: Double

final def isInstanceOf[T0]: Boolean

val lambdas: Array[Double]

final def matrixMult(mat: Matrix, vec: Array[Double], rvec: Array[Double], alpha: Double, beta: Double, trans: Boolean): Unit

val mi: Tensor

val nGates: Int

val nNfs: Int

final def ne(arg0: AnyRef): Boolean

val newA: Array[Double]

val nfs: Int

val nls: Int

final def notify(): Unit

final def notifyAll(): Unit

val numInputFeatures: Int

val numParams: Int

def regularize(): Double

def reset(all: Boolean, slen: Int): Unit

def resetParameters(): Unit

val ri: Matrix

var scale: Array[Double]

val segSize: Int

final def setMatrix(m: Matrix, v: Double = 0.0): Unit

final def setTensor(t: Tensor, v: Double = 0.0): Unit

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

val tmp: Array[Double]

def toString(): String

def train(a: AccessSeq[AbstractInstance], x: Int, modelIterFn: Option[(CoreModel, Int) ⇒ Unit] = None): Nothing

def train(a: AccessSeq[AbstractInstance]): Nothing

def vecSum(vec: Array[Double]): Double

final def wait(): Unit

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

final def wait(arg0: Long): Unit

val weightedActivationPartials: Array[Double]

Inherited from DenseWorker

Inherited from NeuralDenseCrf

Inherited from NeuralStochasticCrfScoring

Inherited from DenseCrf