devolve

Github Page

devolve


alias  HiddenLayer = HiddenNeuron[];

Convenience alias

alias  InputLayer = InputNeuron[];

Convenience alias

alias  OutputLayer = OutputNeuron[];

Convenience alias

abstract class  Neuron;

Abstract class used to describe the neurons which make up the net.

abstract const double  eval();

Get the value of neuron.

abstract const Neuron  clone();

Create a copy of this node. The new node will be unconnected

alias  Connection = std.typecons.Tuple!(double, "weight", Neuron, "neuron").Tuple;

Convenience alias

Connection[]  connections;

Connected nodes

class  HiddenNeuron: devolve.net.network.Neuron;

Class for neurons which exist on the hidden, inner layers of the net.

const double  eval();

Find the value of the neuron by calculating the weighted sum of the connections

const HiddenNeuron  clone();

Create a new copy of the neuron

class  InputNeuron: devolve.net.network.Neuron;

Class for neurons which exist on the input side of the graph

const double  eval();

Returns the input neurons current value

const InputNeuron  clone();

Create a copy of the input neuron with its current value

double  val;

The current value of the input

class  OutputNeuron: devolve.net.network.Neuron;

Class for neurons which return the values

const double  eval();

Evaluate the net by calculating the weighted sum of the connections. Hyperbolic tangent is called on the sum to map it onto the range (-1, 1)

const OutputNeuron  clone();

Create a new copy of the neuron

class  Network;

Genome used with netGA. Represents a simple layered neural network with inputs and outputs as doubles

double[]  opCall(in double[] inputs);

Evaluate this network using the given inputs

const Network  clone();

Preform a deep copy of an existing network

HiddenLayer[]  hiddenLayers;

Array of inner layers

InputLayer  inputLayer;

Layer of input neurons

OutputLayer  outputLayer;

Array of output neurons