public void trainNetwork(
ArrayList<ArrayList<Double>> input, ArrayList<ArrayList<Double>> output) {
if (w1 == null && w2 == null) {
// The +1 refers to the bias
w1 = initializeWeights(input.get(0).size() + 1, hiddenLNo, w1);
w2 = initializeWeights(hiddenLNo + 1, outputLNo, w2);
}
for (int i = 0; i < input.size(); i++) {
Double[] tempOutput = new Double[output.get(i).size()];
tempOutput = output.get(i).toArray(tempOutput);
MinMax(tempOutput);
}
Double[][] tempInput = new Double[input.size()][input.get(0).size()];
Double[][] tempOutput = new Double[output.size()][output.get(0).size()];
for (int i = 0; i < input.size(); i++) {
tempInput[i] = input.get(i).toArray(tempInput[i]);
tempOutput[i] = output.get(i).toArray(tempOutput[i]);
}
for (int j = 0; j < 300; j++) {
for (int i = 0; i < input.size(); i++) {
train(tempInput[i], scale(tempOutput[i], 0.0, 1.0, Dmin, Dmax));
// train(tempInput, tempOutput);
}
// if(lastTrainedWeights.size()==50 &&
// !(isLearning(w1,input.get(0).size() + 1,hiddenLNo) ||
// isLearning(w2,hiddenLNo + 1,outputLNo))){
// System.out.println(j);
// break;
// }
if (lastTrainedWeights.size() == 10) {
lastTrainedWeights.remove();
}
ArrayList<Double[][]> temp = new ArrayList<Double[][]>();
temp.add(w1);
temp.add(w2);
lastTrainedWeights.add(temp);
}
}
