neuralnet

Type Members

1. abstract class BiasedUnitLayer extends MatrixParameterizedLayer with Serializable

   Abstract layer representing a single row of units with a differentiable activation function and some biases (one bias value for each unit).

2. class FullBipartiteConnection extends MatrixParameterizedLayer with Serializable

   Full bipartite graph between two layers with edges labeled by matrix entries of the weight matrix.

3. trait Layer extends VectorSpace[Layer] with Visualizable

   Represents a single layer of a neural network.

4. class LinearUnitLayer extends BiasedUnitLayer

   Layer that does nothing but adding biases to the input

5. abstract class MatrixParameterizedLayer extends Layer with Serializable

   This is a base class for all layers that are parameterized by a rectangular array of double values.

6. class NeuralNet extends NeuralNetLike[NeuralNet] with Serializable

   Base trait for all neural networks.

7. trait NeuralNetLike[+Repr <: NeuralNetLike[Repr]] extends (Mat) ⇒ Mat with Visualizable

   Implementation trait for all subclasses of neural net.

8. class SigmoidUnitLayer extends BiasedUnitLayer with Serializable

   Layer consisting of a single row of sigmoid units.

9. class UnscaledSigmoidUnitLayer extends SigmoidUnitLayer with Serializable

   Little performance optimization: removing unnecessary multiplications with 1