Package edu.cmu.tetrad.search.utils

Class MultiLayerPerceptronDjl

java.lang.Object
edu.cmu.tetrad.search.utils.MultiLayerPerceptronDjl
public class MultiLayerPerceptronDjl extends Object
The MultiLayerPerceptronDjl class provides a customizable implementation of a Multi-Layer Perceptron (MLP) for tasks like regression or classification using the Deep Java Library (DJL). This class allows the user to define the network architecture, including the input dimension, hidden layers, and type of output.

  • Constructor Summary

    Constructors
    Constructor
    Description
    MultiLayerPerceptronDjl(int inputDim, List<Integer> hiddenLayers, String variableType, float inputScale)
    Constructs a MultiLayerPerceptronDjl object with the specified input dimension, hidden layers, variable type, and input scaling factor.

  • Method Summary

    Modifier and Type
    Method
    Description
    ai.djl.ndarray.NDArray
    forward(ai.djl.ndarray.NDManager manager, ai.djl.ndarray.NDArray input)
    Computes the forward pass of the neural network for a given input.
    ai.djl.ndarray.NDManager
    getManager()
    Returns the NDManager used for managing computational resources.

    Methods inherited from class java.lang.Object

    equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

  • Constructor Details

    • MultiLayerPerceptronDjl

      public MultiLayerPerceptronDjl(int inputDim, List<Integer> hiddenLayers, String variableType, float inputScale)
      Constructs a MultiLayerPerceptronDjl object with the specified input dimension, hidden layers, variable type, and input scaling factor. This builds the architecture of a neural network based on provided configurations such as the number of input features, hidden layer specifications, and the output type (e.g., continuous, multinomial, or binary).
      Parameters:
      inputDim - the number of input features or dimensions.
      hiddenLayers - a list of integers defining the number of neurons in each hidden layer.
      variableType - the type of prediction target, such as "continuous", "binary", or "multinomial". For multinomial, it should specify the number of categories as "multinomial,numCategories".
      inputScale - a scaling factor applied to the input data.

  • Method Details

    • forward

      public ai.djl.ndarray.NDArray forward(ai.djl.ndarray.NDManager manager, ai.djl.ndarray.NDArray input) throws ai.djl.translate.TranslateException
      Computes the forward pass of the neural network for a given input.
      Parameters:
      manager - the NDManager used to manage the computational resources.
      input - the input NDArray to process through the neural network.
      Returns:
      the resulting NDArray after the forward pass through the network.
      Throws:
      ai.djl.translate.TranslateException - if there is an issue during computation or data translation.

    • getManager

      public ai.djl.ndarray.NDManager getManager()
      Returns the NDManager used for managing computational resources.
      Returns:
      the NDManager instance.