public void Calculate(double[] inputVector, int iCount,
double[] outputVector /* =NULL */, int oCount /* =0 */,
NNNeuronOutputsList pNeuronOutputs /* =NULL */)
{
var lit = m_Layers.First();
// first layer is imput layer: directly set outputs of all of its neurons
// to the input vector
if (m_Layers.Count > 1)
{
int count = 0;
if (iCount != lit.m_Neurons.Count)
{
return;
}
foreach (var nit in lit.m_Neurons)
{
if (count < iCount)
{
nit.output = inputVector[count];
count++;
}
}
}
//caculate output of next layers
for (int i = 1; i < m_Layers.Count; i++)
{
m_Layers[i].Calculate();
}
// load up output vector with results
if (outputVector != null)
{
lit = m_Layers[m_Layers.Count - 1];
for (int ii = 0; ii < oCount; ii++)
{
outputVector[ii] = lit.m_Neurons[ii].output;
}
}
// load up neuron output values with results
if (pNeuronOutputs != null)
{
// check for first time use (re-use is expected)
pNeuronOutputs.Clear();
// it's empty, so allocate memory for its use
pNeuronOutputs.Capacity = m_Layers.Count;
foreach (NNLayer nnlit in m_Layers)
{
var layerOut = new NNNeuronOutputs(nnlit.m_Neurons.Count);
for (int ii = 0; ii < nnlit.m_Neurons.Count; ++ii)
{
layerOut.Add(nnlit.m_Neurons[ii].output);
}
pNeuronOutputs.Add(layerOut);
}
}
}
/// <summary>
///
/// </summary>
/// <param name="bFromRandomizedPatternSequence"></param>
/// <returns></returns>
/////////////////////////
/// <summary>
/// Get Next Parttern in Parttern List
/// </summary>
/// <param name="iSequenceNum"></param>
/// <param name="bFromRandomizedPatternSequence"></param>
/// <returns></returns>
public void CalculateNeuralNet(double[] inputVector, int count,
double[] outputVector /* =NULL */, int oCount /* =0 */,
NNNeuronOutputsList pNeuronOutputs /* =NULL */,
bool bDistort /* =FALSE */)
{
// wrapper function for neural net's Calculate() function, needed because the NN is a protected member
// waits on the neural net mutex (using the CAutoMutex object, which automatically releases the
// mutex when it goes out of scope) so as to restrict access to one thread at a time
m_Mutexs[0].WaitOne();
{
if (bDistort != false)
{
GenerateDistortionMap(1.0);
ApplyDistortionMap(inputVector);
}
_NN.Calculate(inputVector, count, outputVector, oCount, pNeuronOutputs);
}
m_Mutexs[0].ReleaseMutex();
}
public void Backpropagate(double[] actualOutput, double[] desiredOutput, int count, NNNeuronOutputsList pMemorizedNeuronOutputs)
{
// backpropagates through the neural net
if ((m_Layers.Count >= 2) == false) // there must be at least two layers in the net
{
return;
}
if ((actualOutput == null) || (desiredOutput == null) || (count >= 256))
{
return;
}
// check if it's time for a weight sanity check
m_cBackprops++;
if ((m_cBackprops % 10000) == 0)
{
// every 10000 backprops
PeriodicWeightSanityCheck();
}
// proceed from the last layer to the first, iteratively
// We calculate the last layer separately, and first, since it provides the needed derviative
// (i.e., dErr_wrt_dXnm1) for the previous layers
// nomenclature:
//
// Err is output error of the entire neural net
// Xn is the output vector on the n-th layer
// Xnm1 is the output vector of the previous layer
// Wn is the vector of weights of the n-th layer
// Yn is the activation value of the n-th layer, i.e., the weighted sum of inputs BEFORE the squashing function is applied
// F is the squashing function: Xn = F(Yn)
// F' is the derivative of the squashing function
// Conveniently, for F = tanh, then F'(Yn) = 1 - Xn^2, i.e., the derivative can be calculated from the output, without knowledge of the input
int iSize = m_Layers.Count;
var dErr_wrt_dXlast = new DErrorsList(m_Layers[m_Layers.Count - 1].m_Neurons.Count);
var differentials = new List <DErrorsList>(iSize);
int ii;
// start the process by calculating dErr_wrt_dXn for the last layer.
// for the standard MSE Err function (i.e., 0.5*sumof( (actual-target)^2 ), this differential is simply
// the difference between the target and the actual
for (ii = 0; ii < m_Layers[m_Layers.Count - 1].m_Neurons.Count; ++ii)
{
dErr_wrt_dXlast.Add(actualOutput[ii] - desiredOutput[ii]);
}
// store Xlast and reserve memory for the remaining vectors stored in differentials
for (ii = 0; ii < iSize - 1; ii++)
{
var m_differential = new DErrorsList(m_Layers[ii].m_Neurons.Count);
for (int kk = 0; kk < m_Layers[ii].m_Neurons.Count; kk++)
{
m_differential.Add(0.0);
}
differentials.Add(m_differential);
}
differentials.Add(dErr_wrt_dXlast); // last one
// now iterate through all layers including the last but excluding the first, and ask each of
// them to backpropagate error and adjust their weights, and to return the differential
// dErr_wrt_dXnm1 for use as the input value of dErr_wrt_dXn for the next iterated layer
bool bMemorized = (pMemorizedNeuronOutputs != null);
for (int jj = iSize - 1; jj > 0; jj--)
{
if (bMemorized != false)
{
m_Layers[jj].Backpropagate(differentials[jj], differentials[jj - 1],
pMemorizedNeuronOutputs[jj], pMemorizedNeuronOutputs[jj - 1], m_etaLearningRate);
}
else
{
m_Layers[jj].Backpropagate(differentials[jj], differentials[jj - 1],
null, null, m_etaLearningRate);
}
}
differentials.Clear();
}
