core

Type Members

1. class Dense extends OutputLayer

   

   Fully connected neural net layer.

2. trait Layer extends Node

   

   Trait for proper "layer" in DL4J neural networks and computational graphs.

   Trait for proper "layer" in DL4J neural networks and computational graphs. Compiles out to DL4J layer.

3. trait Node extends AnyRef

   

   Trait for node in DL4J neural networks and computational graphs.

   Trait for node in DL4J neural networks and computational graphs. Nodes are assumed to have inputs and outputs with "shapes."

4. case class Output(isOutput: Boolean, lossFunction: LossFunction) extends Product with Serializable

   

   Trait for output layers in DL4J neural networks and computational graphs.

5. trait OutputLayer extends Layer

   

   Extension of base layer, used to construct a DL4J OutputLayer after compilation.

   Extension of base layer, used to construct a DL4J OutputLayer after compilation. OutputLayer has an output object and the ability to return an OutputLayer version of itself, by providing a loss function.

6. trait Preprocessor extends Node

   

   Trait for preprocessing layers in DL4J neural networks and computational graphs.

   Trait for preprocessing layers in DL4J neural networks and computational graphs. Compiles out to DL4J InputPreProcessor.

7. trait WrapperLayer extends Layer