DistBeliefTrainStyle

Instance Constructors

new DistBeliefTrainStyle(net: Network, algorithm: WeightUpdater, sc: SparkContext, make: ManipulationType[IN, OUT] = new VectorType(), param: DistBeliefCriteria = DistBeliefCriteria())(implicit arg0: ClassTag[IN], arg1: ClassTag[OUT])

net
Network to be trained
algorithm
Weight update algorithm to be applied
sc
A spark context that network will be distributed
make
Input Operation that supervises how to manipulate input as matrices. This also controls how to compute actual network. (default: VectorType)
param
DistBelief-style Training criteria (default: DistBeliefCriteria)

Type Members

type Pair = (IN, OUT)

Training Pair Type
Training Pair Type

Definition Classes
TrainStyle
type Sampler = (Int) ⇒ Seq[OUT]

Sampler Type
Sampler Type

Definition Classes
TrainStyle
implicit class WeightOp extends Serializable

Implicit weight operation
Implicit weight operation

Definition Classes
TrainStyle

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 accCount: Accumulator[Int]

Accumulator variable for counter
Accumulator variable for counter

Attributes
protected
Definition Classes
MultiThreadTrainStyle
val accNet: Accumulator[IndexedSeq[ScalarMatrix]]

Accumulator variable for networks
Accumulator variable for networks

Attributes
protected
Definition Classes
MultiThreadTrainStyle
val algorithm: WeightUpdater

Weight update algorithm to be applied
Weight update algorithm to be applied

Definition Classes
MultiThreadTrainStyle → TrainStyle
final def asInstanceOf[T0]: T0

Definition Classes
Any
def batch(): Unit

Do mini-batch
Do mini-batch

Definition Classes
DistBeliefTrainStyle → MultiThreadTrainStyle → TrainStyle
var batchFlag: ArrayBuffer[Future[Unit]]

Flag for batch : Is Batch remaining?
Flag for batch : Is Batch remaining?

Attributes
protected
var bcNet: Broadcast[Network]

Spark distributed networks
Spark distributed networks

Attributes
protected
Definition Classes
MultiThreadTrainStyle
def clone(): AnyRef

Attributes
protected[java.lang]
Definition Classes
AnyRef
Annotations
@throws( ... )
final def eq(arg0: AnyRef): Boolean

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

Definition Classes
AnyRef → Any
def fetch(iter: Int): Unit

Fetch weights
Fetch weights
iter
current iteration

Definition Classes
DistBeliefTrainStyle → MultiThreadTrainStyle → TrainStyle
var fetchFlag: Future[Unit]

Flag for fetch : Is fetching?
Flag for fetch : Is fetching?

Attributes
protected
def finalize(): Unit

Attributes
protected[java.lang]
Definition Classes
AnyRef
Annotations
@throws( classOf[java.lang.Throwable] )
def foreachTestSet(n: Int)(fn: ((IN, OUT)) ⇒ Unit): Unit

Iterate over given number of test instances
Iterate over given number of test instances
n
number of random sampled instances
fn
iteratee function

Definition Classes
MultiThreadTrainStyle → TrainStyle
final def getClass(): Class[_]

Definition Classes
AnyRef → Any
def hashCode(): Int

Definition Classes
AnyRef → Any
final def isInstanceOf[T0]: Boolean

Definition Classes
Any
def isUpdateFinished: Future[_]

Indicates whether the asynchrononus update is finished or not.
Indicates whether the asynchrononus update is finished or not.
returns
future object of update

Definition Classes
DistBeliefTrainStyle → TrainStyle
val logger: Logger

Logger
Logger

Attributes
protected
Definition Classes
TrainStyle
val make: ManipulationType[IN, OUT]

Input Operation that supervises how to manipulate input as matrices.
Input Operation that supervises how to manipulate input as matrices. This also controls how to compute actual network. (default: VectorType)

Definition Classes
MultiThreadTrainStyle → TrainStyle
final def ne(arg0: AnyRef): Boolean

Definition Classes
AnyRef
var negFraction: Float

Fraction of negative samples
Fraction of negative samples

Attributes
protected
Definition Classes
MultiThreadTrainStyle
var negOutUniverse: RDD[(Long, OUT)]

Negative Sampler
Negative Sampler

Attributes
protected
Definition Classes
MultiThreadTrainStyle
var negPartitioner: RandomEqualPartitioner

Partitioner for negative samples
Partitioner for negative samples

Attributes
protected
Definition Classes
MultiThreadTrainStyle
val net: Network

Network to be trained
Network to be trained

Definition Classes
MultiThreadTrainStyle → TrainStyle
final def notify(): Unit

Definition Classes
AnyRef
final def notifyAll(): Unit

Definition Classes
AnyRef
val param: DistBeliefCriteria

Training criteria (default: SimpleTrainingCriteria)
Training criteria (default: SimpleTrainingCriteria)

Definition Classes
MultiThreadTrainStyle → TrainStyle
val sc: SparkContext

Definition Classes
MultiThreadTrainStyle
def setNegativeTrainingReference(set: RDD[OUT]): Unit

Set negative sampling method.
Set negative sampling method.
set
all training outputs that will be used for negative training

Definition Classes
MultiThreadTrainStyle → TrainStyle
def setNegativeTrainingReference(set: Seq[OUT]): Unit

Set negative sampling method.
Set negative sampling method.
set
all training outputs that will be used for negative training

Definition Classes
MultiThreadTrainStyle → TrainStyle
def setPositiveTrainingReference(set: RDD[(IN, OUT)]): Unit

Set training instances
Set training instances
set
RDD of training set

Definition Classes
MultiThreadTrainStyle → TrainStyle
def setPositiveTrainingReference(set: Seq[(IN, OUT)]): Unit

Set training instances
Set training instances
set
Sequence of training set

Definition Classes
MultiThreadTrainStyle → TrainStyle
def setTestReference(set: RDD[(IN, OUT)]): Unit

Set testing instances
Set testing instances
set
RDD of testing set

Definition Classes
MultiThreadTrainStyle → TrainStyle
def setTestReference(set: Seq[(IN, OUT)]): Unit

Set testing instances
Set testing instances
set
Sequence of testing set

Definition Classes
MultiThreadTrainStyle → TrainStyle
def stopUntilBatchFinished(): Unit

Non-blocking pending, until all assigned batches are finished
Non-blocking pending, until all assigned batches are finished

Definition Classes
DistBeliefTrainStyle → TrainStyle
final def synchronized[T0](arg0: ⇒ T0): T0

Definition Classes
AnyRef
var testSet: RDD[Pair]

Test Set
Test Set

Attributes
protected
Definition Classes
MultiThreadTrainStyle
var testSize: Float

Size of test set
Size of test set

Attributes
protected
Definition Classes
MultiThreadTrainStyle
def toString(): String

Definition Classes
AnyRef → Any
var trainingFraction: Float

Fraction of mini-batch
Fraction of mini-batch

Attributes
protected
Definition Classes
MultiThreadTrainStyle
var trainingSet: RDD[Pair]

Training set
Training set

Attributes
protected
Definition Classes
MultiThreadTrainStyle
def unpersist(): Unit

Unpersist all
Unpersist all

Definition Classes
MultiThreadTrainStyle
def update(iter: Int): Unit

Send update of weights
Send update of weights
iter
current iteration

Definition Classes
DistBeliefTrainStyle → MultiThreadTrainStyle → TrainStyle
var updateFlag: Future[Unit]

Flag for update : Is updating?
Flag for update : Is updating?

Attributes
protected
var validationEpoch: Int

size of training set
size of training set

Definition Classes
TrainStyle
def validationError(): Float

Calculate validation error
Calculate validation error
returns
validation error

Definition Classes
MultiThreadTrainStyle → TrainStyle
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( ... )

class DistBeliefTrainStyle[IN, OUT] extends MultiThreadTrainStyle[IN, OUT]

Instance Constructors

new DistBeliefTrainStyle(net: Network, algorithm: WeightUpdater, sc: SparkContext, make: ManipulationType[IN, OUT] = new VectorType(), param: DistBeliefCriteria = DistBeliefCriteria())(implicit arg0: ClassTag[IN], arg1: ClassTag[OUT])

Type Members

type Pair = (IN, OUT)

type Sampler = (Int) ⇒ Seq[OUT]

implicit class WeightOp extends Serializable

Value Members

final def !=(arg0: Any): Boolean

final def ##(): Int

final def ==(arg0: Any): Boolean

val accCount: Accumulator[Int]

val accNet: Accumulator[IndexedSeq[ScalarMatrix]]

val algorithm: WeightUpdater

final def asInstanceOf[T0]: T0

def batch(): Unit

var batchFlag: ArrayBuffer[Future[Unit]]

var bcNet: Broadcast[Network]

def clone(): AnyRef

final def eq(arg0: AnyRef): Boolean

def equals(arg0: Any): Boolean

def fetch(iter: Int): Unit

var fetchFlag: Future[Unit]

def finalize(): Unit

def foreachTestSet(n: Int)(fn: ((IN, OUT)) ⇒ Unit): Unit

final def getClass(): Class[_]

def hashCode(): Int

final def isInstanceOf[T0]: Boolean

def isUpdateFinished: Future[_]

val logger: Logger

val make: ManipulationType[IN, OUT]

final def ne(arg0: AnyRef): Boolean

var negFraction: Float

var negOutUniverse: RDD[(Long, OUT)]

var negPartitioner: RandomEqualPartitioner

val net: Network

final def notify(): Unit

final def notifyAll(): Unit

val param: DistBeliefCriteria

val sc: SparkContext

def setNegativeTrainingReference(set: RDD[OUT]): Unit

def setNegativeTrainingReference(set: Seq[OUT]): Unit

def setPositiveTrainingReference(set: RDD[(IN, OUT)]): Unit

def setPositiveTrainingReference(set: Seq[(IN, OUT)]): Unit

def setTestReference(set: RDD[(IN, OUT)]): Unit

def setTestReference(set: Seq[(IN, OUT)]): Unit

def stopUntilBatchFinished(): Unit

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

var testSet: RDD[Pair]

var testSize: Float

def toString(): String

var trainingFraction: Float

var trainingSet: RDD[Pair]

def unpersist(): Unit

def update(iter: Int): Unit

var updateFlag: Future[Unit]

var validationEpoch: Int

def validationError(): Float

final def wait(): Unit

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

final def wait(arg0: Long): Unit

Inherited from MultiThreadTrainStyle[IN, OUT]

Inherited from TrainStyle[IN, OUT]

Inherited from Serializable

Inherited from Serializable

Inherited from AnyRef

Inherited from Any

Ungrouped