public class Perceptron {
	private int trainingSetSize = 0;
	private int inputSize = 2; 		// Fixed for now.
	private double inputs[][]; 
	private double desiredOutputs[];// Output is for a network of one neuron.
	private double[] weights; 
	private double learningRate; 	// TODO: Programmer determined. // add as param?
	
	public void initNetwork(double[][] inputs, double[] desiredOutputs) {
		this.inputs = inputs;
		this.trainingSetSize = inputs.length;
		if (this.trainingSetSize == 0) {
			System.out.println("No training data.");
			System.exit(0);
		}
		this.desiredOutputs = desiredOutputs;
		
		this.weights = new double[inputSize];
		//TODO: initialize the weights to small random values in the range ]0.00..0.10]
		this.weights[0] = 1;// TODO: init to small random weights
		this.weights[1] = 1;// TODO: init to small random weights
		 
		this.learningRate = 1000000; //TODO: Set the learning rate to a non-crazy number.
	}
	
	
	//TODO: write this simple function.
	private double sigmoidActivationFunction(double input){
		
	}

	//TODO: write this simple function.
	private double derivative(double activation){
		
	}
	

	//TODO: write this function. Feel free to grab code from the EngineeredPerceptron
	public void trainPerceptronSigmoid(int epochs){

	}
	
	public void printWeights(){
		for (int j = 0; j < inputSize; j++){
			System.out.printf("\nWeight from input %d is: %f", j, weights[j]);
		}
	}
	
	//TODO: write this function. Feel free to grab code from the EngineeredPerceptron. Now that 
	//      you have a sigmoid activation function, you need to determine a non-binary way of 
	//      determining whether your network produced the right output.
	public void testNetworkSigmoid(){

	}
	
	
}