I don't understand back propagation, I just started trying to teach my self about neural netwoks. I believe I unserstand the concept of what a neuron is and how they work. The problem is that with out a way to train the network, the following code that I've written doesn't do much good. I also looked through the tutorials section, but all of the tutorials seem like they are geared towards learning a language(python, java, ruby,etc.) as opposed to learning a topic(cryptology, AI, data compression, etc.) If anybody could give me a thorough explanation of back propagation or any other way to train an ANN I would really appreciate it.

#include<stdio.h>
#include<math.h>
#include<stdlib.h>

int get_input(void);
double sigmoid(double x);


int main()
{
	int nodes_in_input = 0;//This declares the number of nodes in the input layer
	int nodes_in_hidden = 0;//This declares the number of nodes in the hidden layer
	int nodes_in_output = 0;//This declares the number of nodes in the output layer
	int number_of_inputs = 0;//This is equal to the number of data points in the input data file
	int i = 0;
	int j = 0;
	int k = 0;
	double activation = 0;
	
	
	
	printf("How many nodes in the input layer:\n");
	nodes_in_input = get_input();
	printf("How many nodes in the hidden layer:\n");
	nodes_in_hidden = get_input();
	printf("How many nodes in the output layer:\n");
	nodes_in_output = get_input();
	
	double input_layer[nodes_in_input];
	double hidden_layer[nodes_in_hidden];
	double output_layer[nodes_in_output];
	double input_data[];
	double weights[nodes_in_input];
	
	for(i = 0; i<nodes_in_input; i++)
	{
		weights[i] = rand()/(RAND_MAX + 1.0);
	}
	
	for(k=0; k<nodes_in_input; k++)
	{
		double output = 0.0;
		double activation = 0.0;
		
		
		for(j=0; j<number_of_inputs; j++)
		{
			activation = activation + weights[j]*input_data[j];
			output = sigmoid(activation);
		}
		input_layer[i] = output;
	}
	
	for(i=0; i<nodes_in_hidden; i++)
	{
		double output = 0.0;
		double activation = 0.0;
		
		
		for(j=0; j<nodes_in_input; j++)
		{
			activation = activation + weights[j]*input_layer[j];
			sigmoid(activation);
			output = sigmoid(activation);
		}
		hidden_layer[i] = output;
	}
	
	for(i=0; i<nodes_in_output; i++)
	{
		double output = 0.0;
		double activation = 0.0;
		
		
		for(j=0; j<nodes_in_input; j++)
		{
			activation = activation + weights[j]*hidden_layer[j];
			sigmoid(activation);
			output = sigmoid(activation);
		}
		output_layer[i] = output;
	}
	return 0;
}

int get_input(void)
{
	while(1)
	{
		int number = 0;		
		if ( scanf("%d", &number) == 1 && number > 0)
		{
			return  number;		  
		}			  
			printf("Invalid input!\n");
			printf("Please try again\n");
			while (getchar() != '\n'); //Clears the input buffer
			  
	}
}

double sigmoid(double x)
{
	double p = 1.0;
	double output = 0.0;
	output = 1/(1+ exp(-x/p));
	return output;
}

Does nobody understand how back propagation works?!?