static void Test_XOR()
{
NeuralNetwork network = new NeuralNetwork(new int[] { 3, 25, 25, 1 });
for (int i = 0; i < 5000; i++)
{
network.FeedForward(new double[] { 0, 0, 0 });
network.BackProp(new double[] { 0 });
network.FeedForward(new double[] { 0, 0, 1 });
network.BackProp(new double[] { 1 });
network.FeedForward(new double[] { 0, 1, 0 });
network.BackProp(new double[] { 1 });
network.FeedForward(new double[] { 0, 1, 1 });
network.BackProp(new double[] { 0 });
network.FeedForward(new double[] { 1, 0, 0 });
network.BackProp(new double[] { 1 });
network.FeedForward(new double[] { 1, 0, 1 });
network.BackProp(new double[] { 0 });
network.FeedForward(new double[] { 1, 1, 0 });
network.BackProp(new double[] { 0 });
network.FeedForward(new double[] { 1, 1, 1 });
network.BackProp(new double[] { 1 });
}
Console.WriteLine(String.Format("0 0 0 | {0:F3}", network.FeedForward(new double[] { 0, 0, 0 })[0]));
Console.WriteLine(String.Format("0 0 1 | {0:F3}", network.FeedForward(new double[] { 0, 0, 1 })[0]));
Console.WriteLine(String.Format("0 1 0 | {0:F3}", network.FeedForward(new double[] { 0, 1, 0 })[0]));
Console.WriteLine(String.Format("0 1 1 | {0:F3}", network.FeedForward(new double[] { 0, 1, 1 })[0]));
Console.WriteLine(String.Format("1 0 0 | {0:F3}", network.FeedForward(new double[] { 1, 0, 0 })[0]));
Console.WriteLine(String.Format("1 0 1 | {0:F3}", network.FeedForward(new double[] { 1, 0, 1 })[0]));
Console.WriteLine(String.Format("1 1 0 | {0:F3}", network.FeedForward(new double[] { 1, 1, 0 })[0]));
Console.WriteLine(String.Format("1 1 1 | {0:F3}", network.FeedForward(new double[] { 1, 1, 1 })[0]));
Console.WriteLine("\n\n");
//Console.ReadKey();
}
static void Test_Kfold_Iris()
{
NeuralNetwork network = new NeuralNetwork(new int[] { 4, 7, 5, 3 });
List <Tuple <double[], double[]> > input = new List <Tuple <double[], double[]> >();
using (var reader = new StreamReader(@".\input.csv"))
{
reader.ReadLine();
while (!reader.EndOfStream)
{
var line = reader.ReadLine();
var values = line.Split(',');
input.Add(new Tuple <double[], double[]>(new double[] { Double.Parse(values[0], CultureInfo.InvariantCulture),
Double.Parse(values[1], CultureInfo.InvariantCulture),
Double.Parse(values[2], CultureInfo.InvariantCulture),
Double.Parse(values[3], CultureInfo.InvariantCulture) },
MagicSwitch(values[4])));
}
}
Random rng = new Random();
int n = input.Count;
while (n > 1)
{
n--;
int l = rng.Next(n + 1);
Tuple <double[], double[]> value = input[l];
input[l] = input[n];
input[n] = value;
}
int k = 10;
double score = 0;
double[] results;
List <Tuple <double[], double[]> >[] chunks = new List <Tuple <double[], double[]> > [10];
int chunkSize = input.Count() / k;
for (int i = 0; i < k; i++)
{
chunks[i] = input.GetRange(i * chunkSize, chunkSize);
}
for (int j = 0; j < k; j++)
{
Console.WriteLine("Chunk " + (j + 1));
for (int i = 0; i < k; i++)
{
if (i != j)
{
for (int m = 0; m < 5000; m++)
{
foreach (Tuple <double[], double[]> t in chunks[i])
{
network.FeedForward(t.Item1);
network.BackProp(t.Item2);
}
}
}
}
score = 0.0d;
foreach (Tuple <double[], double[]> t in chunks[j])
{
results = network.FeedForward(t.Item1);
List <double> r1 = results.ToList();
List <double> r2 = t.Item2.ToList();
if (r1.IndexOf(r1.Max()) == r2.IndexOf(r2.Max()))
{
score += 1.0d;
}
}
score = score / chunkSize;
Console.WriteLine("Accuracy: " + score * 100 + "%");
}
//double[] result = network.FeedForward(new double[] { 5.8d, 4.0d, 1.2d, 0.2d });
//foreach (double d in result)
//{
// Console.Write(d + " ");
//}
score = 0.0d;
foreach (Tuple <double[], double[]> t in input)
{
results = network.FeedForward(t.Item1);
List <double> r1 = results.ToList();
List <double> r2 = t.Item2.ToList();
if (r1.IndexOf(r1.Max()) == r2.IndexOf(r2.Max()))
{
score += 1.0d;
}
}
score = score / input.Count();
Console.WriteLine("Accuracy: " + score * 100 + "%");
Console.ReadKey();
}
static void Test_Iris()
{
NeuralNetwork network = new NeuralNetwork(new int[] { 4, 7, 5, 3 });
List <Tuple <double[], double[]> > input = new List <Tuple <double[], double[]> >();
using (var reader = new StreamReader(@".\input.csv"))
{
reader.ReadLine();
while (!reader.EndOfStream)
{
var line = reader.ReadLine();
var values = line.Split(',');
input.Add(new Tuple <double[], double[]>(new double[] { Double.Parse(values[0], CultureInfo.InvariantCulture),
Double.Parse(values[1], CultureInfo.InvariantCulture),
Double.Parse(values[2], CultureInfo.InvariantCulture),
Double.Parse(values[3], CultureInfo.InvariantCulture) },
MagicSwitch(values[4])));
}
}
Random rng = new Random();
int n = input.Count;
while (n > 1)
{
n--;
int l = rng.Next(n + 1);
Tuple <double[], double[]> value = input[l];
input[l] = input[n];
input[n] = value;
}
for (int i = 1; i < 5000; i++)
{
foreach (Tuple <double[], double[]> t in input)
{
network.FeedForward(t.Item1);
network.BackProp(t.Item2);
}
}
double[] result = network.FeedForward(new double[] { 5.0, 3.4, 1.6, 0.4 });
Console.Write("Setosa:\t\t");
foreach (double d in result)
{
Console.Write("{0:F3}\t", d);
}
Console.WriteLine();
result = network.FeedForward(new double[] { 6.0, 2.2, 4.0, 1.0 });
Console.Write("Versicolor:\t");
foreach (double d in result)
{
Console.Write("{0:F3}\t", d);
}
Console.WriteLine();
result = network.FeedForward(new double[] { 6.1, 3.0, 4.9, 1.8 });
Console.Write("Virginica:\t");
foreach (double d in result)
{
Console.Write("{0:F3}\t", d);
}
Console.WriteLine("\n");
//Console.ReadKey();
}