randomSeed
TypeScriptget randomSeed(): numberSwift declarationvar randomSeed: IntA seed to generate reproducible results from random operations such as column and row subsampling.
struct · iOS 16.0+ · SDK 27.1
struct FullyConnectedNetworkConfigurationA fully connected network configuration.
get randomSeed(): numberSwift declarationvar randomSeed: IntA seed to generate reproducible results from random operations such as column and row subsampling.
set randomSeed(value: number)Swift declarationvar randomSeed: IntA seed to generate reproducible results from random operations such as column and row subsampling.
get learningRate(): numberSwift declarationvar learningRate: FloatThe learning rate. The learning rate controls how much the model changes when presented with new data. A high learning rate may overshoot when close to a solution, while a low learning rate my take too long to train a good model. Defaults to 0.001.
set learningRate(value: number)Swift declarationvar learningRate: FloatThe learning rate. The learning rate controls how much the model changes when presented with new data. A high learning rate may overshoot when close to a solution, while a low learning rate my take too long to train a good model. Defaults to 0.001.
get hiddenUnitCounts(): Array<number>Swift declarationvar hiddenUnitCounts: [Int]The number of neurons in each hidden layer. Defaults to a single hidden layer with 100 neurons.
set hiddenUnitCounts(value: Array<number>)Swift declarationvar hiddenUnitCounts: [Int]The number of neurons in each hidden layer. Defaults to a single hidden layer with 100 neurons.
get maximumIterations(): numberSwift declarationvar maximumIterations: IntThe maximum number of iterations. More iterations will produce better models as long as there is no over-fitting. Over-fitting happens when the dropout probability is too low or there is not enough training data. - Note: This parameter is only used by the `fitted` method. When using the `update` method it's up to you to decide when to stop.
set maximumIterations(value: number)Swift declarationvar maximumIterations: IntThe maximum number of iterations. More iterations will produce better models as long as there is no over-fitting. Over-fitting happens when the dropout probability is too low or there is not enough training data. - Note: This parameter is only used by the `fitted` method. When using the `update` method it's up to you to decide when to stop.
get dropoutProbability(): numberSwift declarationvar dropoutProbability: FloatThe dropout probability. Dropout layers are placed after fully-connected layers to help prevent over-fitting. Defaults to 0.2.
set dropoutProbability(value: number)Swift declarationvar dropoutProbability: FloatThe dropout probability. Dropout layers are placed after fully-connected layers to help prevent over-fitting. Defaults to 0.2.
get earlyStoppingTolerance(): numberSwift declarationvar earlyStoppingTolerance: DoubleThe early-stopping tolerance. The tolerance is used by the `fitted` method to decide when progress is no longer being made, in which case the training process will stop before the specified maximum number of iterations (known as early stopping). Significant progress happens when the validation loss decreases by at least the tolerance. Defaults to 0.01. - Note: Early stopping only happens when using the `fitted` method with validation data.
set earlyStoppingTolerance(value: number)Swift declarationvar earlyStoppingTolerance: DoubleThe early-stopping tolerance. The tolerance is used by the `fitted` method to decide when progress is no longer being made, in which case the training process will stop before the specified maximum number of iterations (known as early stopping). Significant progress happens when the validation loss decreases by at least the tolerance. Defaults to 0.01. - Note: Early stopping only happens when using the `fitted` method with validation data.
get earlyStopIterationCount(): numberSwift declarationvar earlyStopIterationCount: IntThe number of iterations to use when evaluating whether to stop early. The `fitted` method will stop if no significant progress is made for this many iterations. Significant progress happens when the validation loss decreases by at least `earlyStoppingTolerance`. Defaults to 10. - Note: Early stopping only happens when using the `fitted` method with validation data.
set earlyStopIterationCount(value: number)Swift declarationvar earlyStopIterationCount: IntThe number of iterations to use when evaluating whether to stop early. The `fitted` method will stop if no significant progress is made for this many iterations. Significant progress happens when the validation loss decreases by at least `earlyStoppingTolerance`. Defaults to 10. - Note: Early stopping only happens when using the `fitted` method with validation data.
get batchSize(): numberSwift declarationvar batchSize: IntThe number of examples to use per mini-batch. A larger batch size will speed up computation when training, as long as the batch fits in memory. - Note: This parameter is only used by the `fitted` method. When using the `update` method it's up to you to do batching.
set batchSize(value: number)Swift declarationvar batchSize: IntThe number of examples to use per mini-batch. A larger batch size will speed up computation when training, as long as the batch fits in memory. - Note: This parameter is only used by the `fitted` method. When using the `update` method it's up to you to do batching.
static init(): FullyConnectedNetworkConfigurationSwift declarationinit()Creates a default fully-connected-network configuration.
No declarations were skipped in this snapshot.