Neural Networks in Plain English (Part 2)
This is second part of series , you can find first part here , continuing from where I left , in this article I will explain what is training , and how did I assume 1 as threshold in last article

When we are born , we don’t know things and people around us , our parents , teacher shows example like hey look this is a cat and we first time look at a cat and then understand that cat looks like this , we are given lots of example of how a cat looks and how it is different from other animals . And finally we learn what a cat is and we can easily look at animal and tell whether it is a cat or not . This may look like a simple phenomena .But our brain gets trained multiple times to learn how a cat looks . This is called training
In last article I showed my love for South Indian Foods , this means from my childhood I have eaten different types of food and slowly developed likeness for South Indian food . Technically my brain is trained now , I have seen and eaten South Indian Food many times , that’s why my brain gave higher preference to food type .
This is the exact idea behind perceptron.
In a similar manner when we are building artificial neural network , inputs are given to perceptron one after another and weights get modified according to this equation
For all inputs i,
W(i) = W(i) + l*(Correct Output - Output given by perceptron)*Input(i), where l is the learning rate
I have written whole name equation to make sure any one does not get confused by variable names . This is a very important equation .
So , what is happening ?
Perceptron is adding all the inputs and separating them into 2 categories, those that cause it to fire and those that don’t.
w1x1 + w2x2 = t, where t is the threshold
This is it , in next article I will explain you about what is back propagation and how it is useful .
Thank you for reading .
You can connect me with me in Linkedin
