# MADE A SIMPLE NEURAL NETWORK, BUT NOT SURE IF I DID IT RIGHT

This topic is 2055 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

## Recommended Posts

A simple single unit adaptive network:

The network has 2 inputs, and one output. All are binary. The output
is

1 if W0 *I0 + W1 * I1+ Wb > 0

0 if W0 *I0 + W1 * I1+ Wb <= 0

We want it to learn simple OR: output a 1 if either I0 or I1 is 1.

For Solving this problem i have made this-

#include <iostream>

struct Newron
{
int value;
int weight;
};

int main()
{
Newron input_one, input_two, output;
int bias = 0;

input_one.weight = 1;
input_two.weight = 1;

std::cin >> input_one.value;
std::cin >> input_two.value;

output.value = (input_one.value * input_one.weight) + (input_two.value * input_two.weight) + bias;

if(output.value > 0)
std::cout << 1 << std::endl;
else
std::cout << 0 << std::endl;

return 0;
}


Is it the right way?

If not then how it can be done?

Edited by kazisami

##### Share on other sites

I can't comment, but I'll suggest you a) DON'T MAKE YOUR TITLE LOOK LIKE IT'S SHOUTING, and b) use [ code ][ /code ] (without spaces) tages for code to help with formatting/highlighting.

Good luck!

##### Share on other sites

I can't comment, but I'll suggest you a) DON'T MAKE YOUR TITLE LOOK LIKE IT'S SHOUTING, and b) use [ code ][ /code ] (without spaces) tages for code to help with formatting/highlighting.

Good luck!

Sorry for the title :D And i changed the code formatting, but couldn't be able to change the title.

##### Share on other sites

Hello

This seems to be a good start for me

In fact you are precisely describing (wanting ?) Perceptron model

This is pseudo-code :

class Neuron
{
float [] inputs; // (with : true is 1.0f, false is 0.0f)
float [] weights;

Constructor(integer nbInputs)
{
inputs=new array [nbInputs +1] (+1 is for including the bias). Or getting a reference on an external array.
weights=new array [nbInputs +1]

fill the weights array with float random values between -1 and 1
set 1.0f in the position corresponding to the bias in the inputs array(typically, first or last position)
}

float computeOutput()
{
sum=dotProduct(inputs,weights);
return sum>0.0f;
}

void learn(float desiredOutput,float learningRate) // adaptation of the weights
{
float output=computeOutput();
float deltaOutput=desiredOutput-output; (so -1.0f or 0.0f or 1.0f)

for each position in the arrays
{
weights[i]+=learningRate*inputs[i]*deltaOutput;
}
}

};


Hope it makes sense

for a OR, it should converge with : w1=w2=lambda, and wBias=0. With lambda>0

Good luck

Edited by Tournicoti

##### Share on other sites

Hello

This seems to be a good start for me

In fact you are precisely describing (wanting ?) Perceptron model

This is pseudo-code :

class Neuron
{
float [] inputs; // (with : true is 1.0f, false is 0.0f)
float [] weights; //idem

Constructor(integer nbInputs)
{
inputs=new array [nbInputs +1] (+1 is for including the bias). Or getting a reference on an external array.
weights=new array [nbInputs +1]

fill the weights array with float random values between -1 and 1
set 1.0f in the position corresponding to the bias in the inputs array(typically, first or last position)
}

float computeOutput()
{
sum=dotProduct(inputs,weights);
return sum>0.0f;
}

void learn(float desiredOutput,float learningRate) // adaptation of the weights
{
float output=computeOutput();
float deltaOutput=desiredOutput-output; (so -1.0f or 0.0f or 1.0f)

for each position in the arrays
{
weights[i]+=learningRate*inputs[i]*deltaOutput;
}
}

};


Hope it makes sense

for a OR, it should converge with : w1=lambda, w2=lambda, et wBias=0, lambda>0

Good luck

Thank you very much :D Most of the part is clear to me, but I didn't understood the learningRate argument in the learn() method. Do I have to adjust it so that the network can learn or adjust weight on its own?

##### Share on other sites

The learning rate is just an user-defined constant, here. It's usually a value between 0 and 1.

It's like defining the 'step' of the adaptation.

I would suggest to try low values first  (0.2 ? even less ?) because inputs[i]*deltaOutput is here a huge value : -1 or 0 or 1 !

Afterwards, the goal is to maximize the learning rate so that the learning process is quicker but still accurate enough for the wanted approximation.

Hope I'm still clear (sorry for my english, I'm getting tired )

Bye

Edited by Tournicoti

##### Share on other sites

Don't worry, your english is as awesome as your explanation :D I am getting a hold of this, but I need to experiment a lot before I can understand it clearly. But you answered my question and that's enough for starting experiment without any confusion.

Thanks a lot for your help :D

##### Share on other sites

Edited by Tournicoti

##### Share on other sites

I would like to add another little thing :

What I described is the Perceptron as it was designed by Frank Rosenblatt (original version)

However there's another (very slightly different !) version in which -1 and 1 are used instead of 0 and 1 to encode booleans.

In term of learning efficiency it's better because  :

• with 0 and 1, the weights will evolve only when deltaOutput is not zero and input is not zero.
• with -1 and 1, the weights will always evolves when deltaOutput is not zero.

And this is 'nothing' to change :

• change input encoding (so no more 0 and 1, but -1 and 1)
• change the activation function :  sum>0.0f ? 1.0f : -1.0f

Bye

##### Share on other sites

Thanks again for the help. You know what, I was going to ask you the same thing. I was looking for some other neural networks, then I came across Hopfield network and there is a step when we need to convert the boolean(1 and 0) to 1 and -1. Then I was thinking if this could be done with Perceptron too. And before I could ask, you answered it :)

Edited by kazisami

1. 1
2. 2
3. 3
Rutin
16
4. 4
5. 5

• 14
• 9
• 9
• 9
• 10
• ### Forum Statistics

• Total Topics
632912
• Total Posts
3009199
• ### Who's Online (See full list)

There are no registered users currently online

×