Beginner's Guide to Deep Learning

The boundary is not always accurate when model output is calculated the very first time.
Let's see how can we make the best possible fit function that separates the two classes.

# The best possible fit function 
We want the perfect the boundary that separates the two classes.

The following illustration will help you visualize this:

What did you observe in the illustration above? Are we moving the point closer to the line or the line closer to the misclassified point? 

<p>What did you observe in the illustration above? Are we moving the point closer to the line or the line closer to the misclassified point?</p>


# How to find the best possible fit?
In order to find the best possible boundary, the perceptron algorithm
predicts the output, compares it with the actual output and learns the optimal weights for drawing the best possible fit function that separates the two classes.


## **Predict**&nbsp; the output


Recall the perceptron equation for making a boundary line:
<center>

$w_1x_1 + w_2x_2 + b = 0$

</center>
The prediction is  given by:

$y'$ = { 1 if  $w_1x_1 + w_2x_2 +... w_nx_n + b$ >= 0 and 0 otherwise}


The perceptron trick helps us to find the best possible line that separates the two classes.  

Chapter 1

Perceptron Trick2

The best possible fit function

Quiz

How to find the best possible fit?

Predict the output