private void MainWindow_Loaded(object sender, RoutedEventArgs e)
{
//IOptimize optimizer = new SGD();// SGD();//AdaGrad();// Momentum();// SGD();
net = new TwoLayerNet(784, 50, 10);
net.LoadWidgets("sampleWidget.td");
OnClickImage(sender, e);
}
public async void TestTwoLayerNet()
{
// TrainingData data = TrainingData.Load("sampleWidget.td");
List <Matrix> matrices = await LoadMNISTAsync();
Matrix x_train = matrices[0];
Matrix t_train = matrices[1];
// Matrix x_test = matrices[2];
// Matrix y_test = matrices[3];
int itersNum = 3000;
int batch_size = 100;
IOptimize optimizer = new SGD();// SGD();//AdaGrad();// Momentum();// SGD();
net = new TwoLayerNet(784, 50, 10);
net.LossUpdated += loss => { this.LossUpdated?.Invoke(loss); };
// net.print += Net_print;
Matrix x_batch = new Matrix(batch_size, 784);
Matrix t_batch = new Matrix(batch_size, 10);
int[] indexs = new int[batch_size];
for (int i = 0; i < itersNum; i++)
{
for (int j = 0; j < batch_size; j++)
{
indexs[j] = random.Next(TrainDataCount);
}
for (int row = 0; row < indexs.Length; row++)
{
int index = indexs[row];
// Console.WriteLine(index);
for (int col = 0; col < x_batch.Y; col++)
{
x_batch[row, col] = x_train[index, col];
}
for (int col = 0; col < t_batch.Y; col++)
{
t_batch[row, col] = t_train[index, col];//
}
}
// Matrix[] grads = net.NumericalGradient(x_batch, t_batch);
Matrix[] grads = net.Gradient(x_batch, t_batch);
optimizer.Update(net.Params, grads);
net.Update();
Console.WriteLine($"损失值:{net.Loss(x_batch, t_batch)} ");
if (i % 300 == 0)
{
Console.WriteLine("识别精度计算中...");
double accuracy = net.Accuracy(x_train, t_train);
AccuracyUpdated?.Invoke(accuracy);
accuracies.Add(accuracy);
Console.WriteLine($"识别精度：{accuracy}");
}
}
net.Save("sampleWidget.td");
}
private void Button_Train_Click(object sender, EventArgs e)
{
listBox_NetworkState.Items.Clear();
NeuronNum = Convert.ToInt32(textBox_NeuronNumber.Text);
LearningRate = Convert.ToDouble(textBox_LearningRate.Text);
epoch = Convert.ToInt32(textBox_MaxEpoch.Text);
NetOutput = new double[OutputSize];
Branch = Convert.ToInt32(textBox_branch.Text);
t = new Thread(BackgroundWork);
int DataNumber = InputData.Length / Branch;
double[] NetInput = new double[InputData[0].Length * Branch];
double[] RealOutput = new double[OutputSize];
TrainNetwork = new TwoLayerNet(NetInput.Length, NeuronNum, OutputSize); //網路初始化
for (int k = 0; k < epoch; k++)
//for (int k = 0; k < 1; k++)
{
double errorEachepoch = 0;
//對輸入進來的482筆資料做訓練,DataNumber = 482
for (int i = 0; i < DataNumber; i++)
//for (int i = 0; i < 3; i++)
{
//處理要輸入進網路的資料NetInput
for (int j = 0; j < Branch; j++)
{
NetInput[0 + 4 * j] = InputData[i * Branch + j][0];
NetInput[1 + 4 * j] = InputData[i * Branch + j][1];
NetInput[2 + 4 * j] = InputData[i * Branch + j][2];
NetInput[3 + 4 * j] = InputData[i * Branch + j][3];
}
//處理該筆資料的RealOutput
if (DataOutput[i * Branch] == 1)
{
RealOutput[0] = 1;
RealOutput[1] = 0;
}
else
{
RealOutput[0] = 0;
RealOutput[1] = 1;
}
//進行forward
NetOutput = TrainNetwork.forward(NetInput);
TrainNetwork.backward(LearningRate, RealOutput, NetInput);
errorEachepoch += TrainNetwork.loss(NetOutput, RealOutput);
//listBox_NetworkState.Items.Add(TrainNetwork.loss(NetOutput, RealOutput));
}
errorEachepoch = errorEachepoch / DataNumber;
listBox_NetworkState.Items.Add(k + " : " + errorEachepoch);
/*
* listBox_NetworkState.Items.Add("Weight : ");
* for (int i = 0; i < TrainNetwork.Weight1.Length; i++)
* {
* string weight_col = "";
* for (int j = 0; j < TrainNetwork.Weight1[0].Length; j++)
* {
* weight_col += TrainNetwork.Weight1[i][j].ToString("f2") + "\t";
* }
* listBox_NetworkState.Items.Add(weight_col);
* }
*
* listBox_NetworkState.Items.Add("Bias : ");
* for (int i = 0; i < TrainNetwork.Bias1.Length; i++)
* {
* listBox_NetworkState.Items.Add(TrainNetwork.Bias1[i].ToString("f2"));
* }
*
* //NetOutput = TrainNetwork.forward(NetInput);
*
*
* listBox_NetworkState.Items.Add("Sum : " + TestSum);
*
* listBox_NetworkState.Items.Add("loss : ");
* listBox_NetworkState.Items.Add(TrainNetwork.loss(NetInput, RealOutput));
* TrainNetwork.backward(LearningRate, RealOutput, NetInput);
*/
}
}
