Interface Optimizer<T>

Type Parameters:
T - the data type.
All Known Implementing Classes:
SGD
public interface Optimizer<T>
Interface for optimization algorithms used to train neural networks. Examples: SGD, Adam, RMSProp.

  • Method Summary

    Modifier and Type
    Method
    Description
    void
    step(Map<String, GraphNode<T>> parameters)
    Updates the parameters based on their internal gradients.
    default void
    zeroGrad(Map<String, GraphNode<T>> parameters)
    Resets gradients to zero.

  • Method Details

    • step

      void step(Map<String, GraphNode<T>> parameters)
      Updates the parameters based on their internal gradients.
      Parameters:
      parameters - valid map of parameters to update.

    • zeroGrad

      default void zeroGrad(Map<String, GraphNode<T>> parameters)
      Resets gradients to zero.
      Parameters:
      parameters - map of parameters to reset.