public void run() { var dictionaryInput = new LetterDataTextFileReader(); var floatLetterInput = dictionaryInput.ConvertTextForFloatArray(@"C:\Users\BartlomiejLeja\source\repos\PerceptronLerning\PerceptronLerning\letterLearningData.txt"); var net = new BackpropagationNeuralNetwork(new int[] { 35, 35, 35, 20 }); //intiilize network 3 input (0 0 0) 2 hidden layers with 25 neurons 1 output (1) // Itterate 5000 times and train each possible output // 5000 * 8 = 40000 traning operations for (int i = 0; i < 5000; i++) { for (int j = 0; j < floatLetterInput.Count; j++) { net.FeedForward(floatLetterInput[j].LetterPattern); net.BackProp(floatLetterInput[j].LetterResult); } } net.TestMethod(new float[] { 0, 1, 1, 1, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 0, 0, 0, 1, 1, 0, 0, 0, 1 }, 'A'); net.TestMethod(new float[] { 1, 1, 1, 1, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 0 }, 'B'); }
public void run() { var dictionaryInput = new LetterDataTextFileReader(); var letterInput = dictionaryInput.ConvertTextForArray(@"C:\Users\BartlomiejLeja\source\repos\PerceptronLerning\PerceptronLerning\letterLearningData.txt"); var oneLayerNetworkTest = new OneLayerNetwork(letterInput, 0.5); }