Package ciir.umass.edu.learning.neuralnet

Class Neuron

  • Direct Known Subclasses:
    ListNeuron
    public class Neuron
extends java.lang.Object
    Author:
    vdang This class implements individual neurons in the network.

    • Field Detail

      • momentum

        public static double momentum

      • learningRate

        public static double learningRate

      • output

        protected double output

      • outputs

        protected java.util.List<java.lang.Double> outputs

      • delta_i

        protected double delta_i

      • deltas_j

        protected double[] deltas_j

      • inLinks

        protected java.util.List<Synapse> inLinks

      • outLinks

        protected java.util.List<Synapse> outLinks

    • Constructor Detail

      • Neuron

        public Neuron()

    • Method Detail

      • getOutput

        public double getOutput()

      • getOutput

        public double getOutput​(int k)

      • getInLinks

        public java.util.List<Synapse> getInLinks()

      • getOutLinks

        public java.util.List<Synapse> getOutLinks()

      • setOutput

        public void setOutput​(double output)

      • addOutput

        public void addOutput​(double output)

      • computeOutput

        public void computeOutput()

      • computeOutput

        public void computeOutput​(int i)

      • clearOutputs

        public void clearOutputs()

      • computeDelta

        public void computeDelta​(PropParameter param)
        Compute delta for neurons in the output layer. ONLY for neurons in the output layer.
        Parameters:
        targetValue -

      • updateDelta

        public void updateDelta​(PropParameter param)
        Update delta from neurons in the next layer (back-propagate)

      • updateWeight

        public void updateWeight​(PropParameter param)
        Update weights of incoming links.