Well having issues with this other program here...

Well i wanted to attempt to model a basic perceptron in C++.

I know this isn't right or nothing, ha, but i gave it a go anyways.

I get all sorts of segfaults throughout, but here's the program:

Code:

#include <iostream>
#include <windows.h>

using namespace std;

struct ending
{
	int input;
};

struct perceptron
{
	ending *ends[8];
	int value;
};

int ops[7];

//Functions
void returnEnds(ending *e[8], int a[8]);
bool learn(perceptron *p, int ops[7], int result);
int createResult(int a[8]);
int operations(int input1, int input2, int op);
void displayEq(perceptron *p, int ops[]);

int main()
{
	//Perceptron (single) AI Self-Learning Program
	perceptron *p;
	p = new perceptron;
	ending *ends[8];
	int a[8];
	int dResult;
	dResult = createResult(a);
	//initialize
	ops[0] = -1; //being used as a flag
	for(int i=0; i<8; i++)
	{
		ends[i] = new ending;
		a[i] = (i*2);
		ends[i]->input = a[i];
	}
	//Display
	cout<<"Target Number: "<<dResult<<endl;
	cout<<"The program will now try and figure out how to achieve this number and constantly"
		<<"compare it's own feedback, comparing it to the result, and hopefully learn the process."<<endl<<endl;
	//Start AI
	while(learn(p, ops, dResult) == false)
	{
		returnEnds(ends, a);
		for(int i=0; i<8; i++)
		{
			p->ends[i] = ends[i];
		}
	}
}

void returnEnds(ending *e[8], int a[8])
{
	for(int i=0; i<8; i++)
	{
		e[i]->input = a[i];
	}
}

int createResult(int a[8])
{
	return(a[0] + a[1] + a[2] - a[3] + a[4] - a[5] + a[6] - a[7]);
}

int operations(int input1, int input2, int op)
{
	int ret=0;
	if(op == 0)
	{
		//addition
		ret = input1 + input2;
	}else if(op == 1)
	{
		//subtraction
		ret = input1 - input2;
	}
	else{
		ret = 0;
	}
	return ret;
}

bool learn(perceptron *p, int ops[7], int result)
{
	if(ops[0] == -1)
	{
		if(result <= 0)
		{
			for(int i=0; i<7; i++)
			{
				ops[i] = 1;
			}
		}else if(result >0)
		{
			for(int i=0; i<7; i++)
			{
				ops[i] = 0;
			}
		}
	}
	int value=0;
	for(int i=0; i<7; i++)
	{
		value += (operations(p->ends[i]->input, p->ends[i+1]->input, ops[i]));
	}
	cout<<"Target: "<<result<<endl;
	cout<<"Value:  "<<value<<endl;
	cout<<"Approach: ";
	displayEq(p, ops);
	cout<<endl;
	Sleep(150);
	if(value > result)
	{
		//Decrease
	}else if(value < result)
	{
		//Increase
	}else if(value == result)
	{
		//Did what it is supposed to, return true
		return true;
	}
	return false;
}

void displayEq(perceptron *p, int ops[])
{
	int j;
	for(int i=0; i<8; i++)
	{
		if(i >= 7)
		{
			j = 6;
		}else{
			j = i;
		}
		cout<<p->ends[i]->input<<" ";
		if(ops[j] == 0)
		{
			//addition
			cout<<"+";
		}else if(ops[j] == 1)
		{
			//subtraction
			cout<<"-";
		}else{
			cout<<" ";
		}
		cout<<" ";
	}
}

Danke!

First, you haven't allocated memory for the ending pointers when you pass the perceptron object to the "learn" function.

Second, I'm having a hard time understanding how this is a perceptron. Normally, you'd use a perceptron for data classification tasks, and I have heard that except for XOR, you can implement boolean relations with a perceptron, but whatever it's for, there are three things you always need no matter what and I don't see that here.

You need
(1) a weight vector
(2) a feature vector
(3) a threshold.

You get the vector product of the weight vector and feature vector, sum over it, compare it to the threshold (which in your case would be dResult), and if the sum is greater, you decrement the weight vectors corresponding to the features and if the sum is less, you increment the weight vectors. You iterate over this until you "converge" or you the marginal increase in accuracy is minimal.

Ah... i see

Well this was just a (sad) attempt on my part