Class

breeze.optimize.AdaptiveGradientDescent

L1Regularization

Related Doc: package AdaptiveGradientDescent

Permalink

class L1Regularization[T] extends StochasticGradientDescent[T]

Implements the L1 regularization update.

Each step is:

x_{t+1}i = sign(x_{t,i} - eta/s_i * g_ti) * (abs(x_ti - eta/s_ti * g_ti) - lambda * eta /s_ti))_+

where g_ti is the gradient and s_ti = \sqrt(\sum_t'{t} g_ti2)

Linear Supertypes
StochasticGradientDescent[T], FirstOrderMinimizer[T, StochasticDiffFunction[T]], SerializableLogging, Serializable, Serializable, Minimizer[T, StochasticDiffFunction[T]], AnyRef, Any
Ordering
  1. Alphabetic
  2. By Inheritance
Inherited
  1. L1Regularization
  2. StochasticGradientDescent
  3. FirstOrderMinimizer
  4. SerializableLogging
  5. Serializable
  6. Serializable
  7. Minimizer
  8. AnyRef
  9. Any
  1. Hide All
  2. Show All
Visibility
  1. Public
  2. All

Instance Constructors

  1. new L1Regularization(lambda: Double = 1.0, delta: Double = 1E-5, eta: Double = 4, maxIter: Int = 100)(implicit space: MutableFiniteCoordinateField[T, _, Double], rand: RandBasis = Rand)

    Permalink

Type Members

  1. case class History(sumOfSquaredGradients: T) extends Product with Serializable

    Permalink

  2. type State = FirstOrderMinimizer.State[T, Info, History]

    Permalink
    Definition Classes
    FirstOrderMinimizer

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 adjust(newX: T, newGrad: T, newVal: Double): (Double, T)

    Permalink
    Attributes
    protected
    Definition Classes
    L1Regularization → FirstOrderMinimizer

  5. def adjustFunction(f: StochasticDiffFunction[T]): StochasticDiffFunction[T]

    Permalink
    Attributes
    protected
    Definition Classes
    FirstOrderMinimizer

  6. final def asInstanceOf[T0]: T0

    Permalink
    Definition Classes
    Any

  7. def calculateObjective(f: StochasticDiffFunction[T], x: T, history: History): (Double, T)

    Permalink
    Attributes
    protected
    Definition Classes
    FirstOrderMinimizer

  8. def chooseDescentDirection(state: State, fn: StochasticDiffFunction[T]): T

    Permalink
    Attributes
    protected
    Definition Classes
    StochasticGradientDescent → FirstOrderMinimizer

  9. def clone(): AnyRef

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

  10. val convergenceCheck: ConvergenceCheck[T]

    Permalink
    Definition Classes
    FirstOrderMinimizer

  11. val defaultStepSize: Double

    Permalink
    Definition Classes
    StochasticGradientDescent

  12. def determineStepSize(state: State, f: StochasticDiffFunction[T], dir: T): Double

    Permalink

    Choose a step size scale for this iteration.

    Choose a step size scale for this iteration.

    Default is eta / math.pow(state.iter + 1,2.0 / 3.0)

    Definition Classes
    L1Regularization → StochasticGradientDescent → FirstOrderMinimizer

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

    Permalink
    Definition Classes
    AnyRef

  14. def equals(arg0: Any): Boolean

    Permalink
    Definition Classes
    AnyRef → Any

  15. def finalize(): Unit

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

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

    Permalink
    Definition Classes
    AnyRef → Any

  17. def hashCode(): Int

    Permalink
    Definition Classes
    AnyRef → Any

  18. def infiniteIterations(f: StochasticDiffFunction[T], state: State): Iterator[State]

    Permalink
    Definition Classes
    FirstOrderMinimizer

  19. def initialHistory(f: StochasticDiffFunction[T], init: T): History

    Permalink
    Definition Classes
    L1Regularization → FirstOrderMinimizer

  20. def initialState(f: StochasticDiffFunction[T], init: T): State

    Permalink
    Attributes
    protected
    Definition Classes
    FirstOrderMinimizer

  21. final def isInstanceOf[T0]: Boolean

    Permalink
    Definition Classes
    Any

  22. def iterations(f: StochasticDiffFunction[T], init: T): Iterator[State]

    Permalink
    Definition Classes
    FirstOrderMinimizer

  23. val lambda: Double

    Permalink

  24. def logger: LazyLogger

    Permalink
    Attributes
    protected
    Definition Classes
    SerializableLogging

  25. val maxIter: Int

    Permalink
    Definition Classes
    StochasticGradientDescent

  26. def minimize(f: StochasticDiffFunction[T], init: T): T

    Permalink
    Definition Classes
    FirstOrderMinimizer → Minimizer

  27. def minimizeAndReturnState(f: StochasticDiffFunction[T], init: T): State

    Permalink
    Definition Classes
    FirstOrderMinimizer

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

    Permalink
    Definition Classes
    AnyRef

  29. final def notify(): Unit

    Permalink
    Definition Classes
    AnyRef

  30. final def notifyAll(): Unit

    Permalink
    Definition Classes
    AnyRef

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

    Permalink
    Definition Classes
    AnyRef

  32. def takeStep(state: State, dir: T, stepSize: Double): T

    Permalink

    Projects the vector x onto whatever ball is needed.

    Projects the vector x onto whatever ball is needed. Can also incorporate regularization, or whatever.

    Default just takes a step

    Attributes
    protected
    Definition Classes
    L1Regularization → StochasticGradientDescent → FirstOrderMinimizer

  33. def toString(): String

    Permalink
    Definition Classes
    AnyRef → Any

  34. def updateHistory(newX: T, newGrad: T, newValue: Double, f: StochasticDiffFunction[T], oldState: State): History

    Permalink
    Definition Classes
    L1Regularization → FirstOrderMinimizer

  35. implicit val vspace: NormedModule[T, Double]

    Permalink
    Attributes
    protected
    Definition Classes
    StochasticGradientDescent

  36. final def wait(): Unit

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

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

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

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

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

Inherited from StochasticGradientDescent[T]

Inherited from FirstOrderMinimizer[T, StochasticDiffFunction[T]]

Inherited from SerializableLogging

Inherited from Serializable

Inherited from Serializable

Inherited from Minimizer[T, StochasticDiffFunction[T]]

Inherited from AnyRef

Inherited from Any

Ungrouped