Object

com.enriquegrodrigo.spark.crowd.methods

Glad

Related Doc: package methods

Permalink

object Glad

Provides functions for transforming an annotation dataset into a standard label dataset using the Glad algorithm

This algorithm only works with com.enriquegrodrigo.spark.crowd.types.BinaryAnnotation datasets

Example: 

  1. result: GladModel = Glad(dataset)
See also

Whitehill, Jacob, et al. "Whose vote should count more: Optimal integration of labels from labelers of unknown expertise." Advances in neural information processing systems. 2009.

Linear Supertypes
AnyRef, Any
Ordering
  1. Alphabetic
  2. By Inheritance
Inherited
  1. Glad
  2. AnyRef
  3. Any
  1. Hide All
  2. Show All
Visibility
  1. Public
  2. All

Type Members

  1. class GladAlphaAggregator extends Aggregator[GladPartial, GladAlphaAggregatorBuffer, Double]

    Permalink

    Aggregator for the precision of annotators

    Aggregator for the precision of annotators

    Version

    0.1

  2. case class GladAlphaAggregatorBuffer(agg: Double, alpha: Double) extends Product with Serializable

    Permalink

    Buffer for the alpha aggregator

    Buffer for the alpha aggregator

    Version

    0.1

  3. case class GladAlphas(annotator: Long, alpha: Double) extends Product with Serializable

    Permalink

    Class that storage the reliability for an annotator

    Class that storage the reliability for an annotator

    Version

    0.1

  4. case class GladAnnotatorPrecision(annotator: Long, alpha: Int) extends Product with Serializable

    Permalink

    Glad Annotator precision estimation

    Glad Annotator precision estimation

    Version

    0.1

  5. class GladBetaAggregator extends Aggregator[GladPartial, GladBetaAggregatorBuffer, Double]

    Permalink

    Aggregator for the difficulty of each example

    Aggregator for the difficulty of each example

    Version

    0.1

  6. case class GladBetaAggregatorBuffer(agg: Double, beta: Double) extends Product with Serializable

    Permalink

    Buffer for the beta aggregator

    Buffer for the beta aggregator

    Version

    0.1

  7. class GladEAggregator extends Aggregator[GladPartial, GladEAggregatorBuffer, Double]

    Permalink

    Aggregator for the E step of the EM algorithm

    Aggregator for the E step of the EM algorithm

    Version

    0.1

  8. case class GladEAggregatorBuffer(aggVect: Seq[Double]) extends Product with Serializable

    Permalink

    Buffer for the E step aggregator

    Buffer for the E step aggregator

    Version

    0.1

  9. class GladLogLikelihoodAggregator extends Aggregator[GladPartial, GladLogLikelihoodAggregatorBuffer, Double]

    Permalink

    Aggregator for the Likelihood estimation of the EM algorithm

    Aggregator for the Likelihood estimation of the EM algorithm

    Version

    0.1

  10. case class GladLogLikelihoodAggregatorBuffer(agg: Double, classProb: Double) extends Product with Serializable

    Permalink

    Buffer for the likelihood aggregator

    Buffer for the likelihood aggregator

    Version

    0.1

  11. case class GladParams(alpha: Array[Double], w: Array[Double]) extends Product with Serializable

    Permalink

    Case class for storing Glad model parameters

    Case class for storing Glad model parameters

    Version

    0.1

  12. case class GladPartial(example: Long, annotator: Int, value: Int, est: Double, beta: Double) extends Product with Serializable

    Permalink

    Case class for storing annotations with class estimations and beta

    Case class for storing annotations with class estimations and beta

    Version

    0.1

  13. case class GladPartialModel(dataset: Dataset[GladPartial], params: Broadcast[GladParams], logLikelihood: Double, improvement: Double, nAnnotators: Int) extends Product with Serializable

    Permalink

    *************************************************

Value Members

  1. final def !=(arg0: Any): Boolean

    Permalink
    Definition Classes
    AnyRef → Any

  2. final def ##(): Int

    Permalink
    Definition Classes
    AnyRef → Any

  3. final def ==(arg0: Any): Boolean

    Permalink
    Definition Classes
    AnyRef → Any

  4. def apply(dataset: Dataset[BinaryAnnotation], eMIters: Int = 1, eMThreshold: Double = 0.001, gradIters: Int = 3, gradThreshold: Double = 0.5, gradLearningRate: Double = 0.1, alphaPrior: Double = 1, betaPrior: Double = 1): GladModel

    Permalink

    Apply the Glad Algorithm.

    Apply the Glad Algorithm.

    dataset

    The dataset over which the algorithm will execute.

    eMIters

    Number of iterations for the EM algorithm

    eMThreshold

    LogLikelihood variability threshold for the EM algorithm

    gradIters

    Maximum number of iterations for the GradientDescent algorithm

    gradThreshold

    Threshold for the log likelihood variability for the gradient descent algorithm

    gradLearningRate

    Learning rate for the gradient descent algorithm

    alphaPrior

    First value for all alpha parameters

    betaPrior

    First value for all beta parameters

    Version

    0.1

  5. final def asInstanceOf[T0]: T0

    Permalink
    Definition Classes
    Any

  6. def clone(): AnyRef

    Permalink
    Attributes
    protected[java.lang]
    Definition Classes
    AnyRef
    Annotations
    @throws( ... )

  7. def eStep(model: GladPartialModel): GladPartialModel

    Permalink

    The E step from the EM algorithm

    The E step from the EM algorithm

    Version

    0.1

  8. final def eq(arg0: AnyRef): Boolean

    Permalink
    Definition Classes
    AnyRef

  9. def equals(arg0: Any): Boolean

    Permalink
    Definition Classes
    AnyRef → Any

  10. def finalize(): Unit

    Permalink
    Attributes
    protected[java.lang]
    Definition Classes
    AnyRef
    Annotations
    @throws( classOf[java.lang.Throwable] )

  11. final def getClass(): Class[_]

    Permalink
    Definition Classes
    AnyRef → Any

  12. def hashCode(): Int

    Permalink
    Definition Classes
    AnyRef → Any

  13. def initialization(dataset: Dataset[BinaryAnnotation], alphaPrior: Double, betaPrior: Double): GladPartialModel

    Permalink

    Initialization of the parameters of the algorithm.

    Initialization of the parameters of the algorithm. The first label estimation is done using majority voting

    Version

    0.1

  14. final def isInstanceOf[T0]: Boolean

    Permalink
    Definition Classes
    Any

  15. def logLikelihood(model: GladPartialModel): GladPartialModel

    Permalink

    Neg-logLikelihood calculation

    Neg-logLikelihood calculation

    Version

    0.1

  16. def mStep(model: GladPartialModel, maxGradIters: Int, thresholdGrad: Double, learningRate: Double): GladPartialModel

    Permalink

    The M step from the EM algorithm

    The M step from the EM algorithm

    Version

    0.1

  17. final def ne(arg0: AnyRef): Boolean

    Permalink
    Definition Classes
    AnyRef

  18. final def notify(): Unit

    Permalink
    Definition Classes
    AnyRef

  19. final def notifyAll(): Unit

    Permalink
    Definition Classes
    AnyRef

  20. def step(maxGradIters: Int, thresholdGrad: Double, learningRate: Double)(model: GladPartialModel, i: Int): GladPartialModel

    Permalink

    Full step of the EM algorithm

    Full step of the EM algorithm

    Version

    0.1

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

    Permalink
    Definition Classes
    AnyRef

  22. def toString(): String

    Permalink
    Definition Classes
    AnyRef → Any

  23. final def wait(): Unit

    Permalink
    Definition Classes
    AnyRef
    Annotations
    @throws( ... )

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

    Permalink
    Definition Classes
    AnyRef
    Annotations
    @throws( ... )

  25. final def wait(arg0: Long): Unit

    Permalink
    Definition Classes
    AnyRef
    Annotations
    @throws( ... )

Inherited from AnyRef

Inherited from Any

Ungrouped