private void UpdateWeights(TPM.TPM tpm)
{
var output = tpm.OutputLayer.GetOutput();
foreach (var neuron in tpm.InputLayer.Neurons)
{
foreach (var synapse in neuron.TargetSynapses)
{
if (output != synapse.TargetNeuron.Output) //may be problem
{
continue;
}
var newWeight = synapse.Weight;
newWeight += neuron.Input * output;
if (newWeight > configuration.WeightRange)
{
newWeight = configuration.WeightRange;
}
if (newWeight < configuration.WeightRange * -1)
{
newWeight = configuration.WeightRange * -1;
}
synapse.Weight = newWeight;
}
}
}
public void SyncTPM(TPM.TPM m1, TPM.TPM m2)
{
for (int i = 0; i < int.MaxValue; i++)
{
var weightsOld1 = m1.InputLayer.GetTargetWeights();
var weightsold2 = m2.InputLayer.GetTargetWeights();
var rand = new Random(DateTime.Now.Millisecond);
var input = new List <int>();
for (int j = 0; j < m1.InputLayer.Neurons.Count; j++)
{
var r = rand.Next(-1, 2);
while (r == 0)
{
r = rand.Next(-1, 2);
}
input.Add(r);
}
m1.Run(input.ToArray());
m2.Run(input.ToArray());
if (m1.OutputLayer.GetOutput() == m2.OutputLayer.GetOutput())
{
UpdateWeights(m1);
UpdateWeights(m2);
}
var weights1 = m1.InputLayer.GetTargetWeights();
var weights2 = m2.InputLayer.GetTargetWeights();
var success = true;
for (var k = 0; k < weights1.Count; k++)
{
if (weights1[k] != weights2[k])
{
success = false;
break;
}
}
if (success)
{
throw new Exception();
}
}
}
private void UpdateWeights(TPM.TPM tpm)
{
var output = tpm.OutputLayer.GetOutput();
foreach (var neuron in tpm.InputLayer.Neurons)
{
foreach (var synapse in neuron.TargetSynapses)
{
if (output != synapse.TargetNeuron.Output) //may be problem
continue;
var newWeight = synapse.Weight;
newWeight += neuron.Input * output;
if (newWeight > configuration.WeightRange)
newWeight = configuration.WeightRange;
if (newWeight < configuration.WeightRange * -1)
newWeight = configuration.WeightRange * -1;
synapse.Weight = newWeight;
}
}
}
public void SyncTPM(TPM.TPM m1 , TPM.TPM m2)
{
for (int i = 0; i < int.MaxValue; i++)
{
var weightsOld1 = m1.InputLayer.GetTargetWeights();
var weightsold2 = m2.InputLayer.GetTargetWeights();
var rand = new Random(DateTime.Now.Millisecond);
var input = new List<int>();
for (int j = 0; j < m1.InputLayer.Neurons.Count; j++)
{
var r = rand.Next(-1, 2);
while (r == 0)
{
r = rand.Next(-1, 2);
}
input.Add(r);
}
m1.Run(input.ToArray());
m2.Run(input.ToArray());
if (m1.OutputLayer.GetOutput() == m2.OutputLayer.GetOutput()){
UpdateWeights(m1);
UpdateWeights(m2);
}
var weights1 = m1.InputLayer.GetTargetWeights();
var weights2 = m2.InputLayer.GetTargetWeights();
var success = true;
for (var k = 0; k < weights1.Count; k++)
{
if (weights1[k] != weights2[k])
{
success = false;
break;
}
}
if (success)
{
throw new Exception();
}
}
}
