abstractnn

Type Members

1. abstract class AbstractCriterion[A <: Activity, B <: Activity, T] extends Serializable

   

   AbstractCriterion is an abstract class the concrete criterion should extend.

   AbstractCriterion is an abstract class the concrete criterion should extend. Criterions are helpful to train a neural network. Given an input and a target, they compute the gradient according to a loss function.

   It provides some important method such as forward, backward, updateOutput, updateGradInput frequently used as a criteria. Some of them need to be override in a concrete criterion class.

   A

   represents the input type of the criterion, which an be abstract type Activity, or concrete type Tensor or Table

   B

   represents the output type of the criterion

   T

   The numeric type in the criterion, usually which are Float or Double

2. abstract class AbstractModule[A <: Activity, B <: Activity, T] extends Serializable with InferShape

   

   Module is the basic component of a neural network.

   Module is the basic component of a neural network. It forward activities and backward gradients. Modules can connect to others to construct a complex neural network.

   A

   Input data type

   B

   Output data type

   T

   The numeric type in this module parameters.

3. trait Activity extends AnyRef

   

   Activity is a trait which represents the concept of neural input within neural networks.

   Activity is a trait which represents the concept of neural input within neural networks. For now, two type of input are supported and extending this trait, which are Tensor and Table.

4. sealed trait DataFormat extends AnyRef

   

   DataFormat are used to specify the data format of the input and output data when data is 2-D images.

5. class EmptyGradInput extends Activity with Serializable

   

   Sometimes a module may not have gradInput in the backward(e.g.

   Sometimes a module may not have gradInput in the backward(e.g. some operation layer or stopGradient in a Graph). This is allowed when the gradInput is not used anywhere.

   In such case, the gradInput of the module should be marked as EmptyGradInput. This class make sure an error will happen when user try to use such gradInput.

6. trait IdentityOutputShape extends InferShape

   

7. trait InferShape extends AnyRef

   

8. trait Initializable extends AnyRef

   

   The trait that contains setInitMethod

9. class InvalidLayer extends RuntimeException

   

10. abstract class TensorCriterion[T] extends AbstractCriterion[Tensor[T], Tensor[T], T]

    

    TensorCriterion is an abstract sub-class of AbstractCriterion, whose input and output type both are Tensor.

    TensorCriterion is an abstract sub-class of AbstractCriterion, whose input and output type both are Tensor.

    T

    The numeric type in the criterion, usually which are Float or Double

11. abstract class TensorModule[T] extends AbstractModule[Tensor[T], Tensor[T], T]

    

    TensorModule is an abstract sub-class of AbstractModule, whose input and output type both are Tensor.

    TensorModule is an abstract sub-class of AbstractModule, whose input and output type both are Tensor.

    T

    The numeric type in this module parameters