Friday, September 4, 2020

Perceptrons- Single layer and Multiple Layer Artificial Neural Network

 


  What is a Perceptron ?

   Perceptron  is  a   learning  algorithm  used in  machine  learning  which is  used

   to  do  binary  classification  and  decide  the input  category and provide output

   according  to  the  input  given.

 

 Single Layer  Perceptron :

 It is the simplest  type  of  Artificial Neural  Network.It is calculated by  computing

 the sum of all input vector multiplied  by  corresponding  weights.The value which

is displayed in the output  will  be  the  value of  the  Activation  function.

Single Layer Perceptron Sums up all  the  weighted values and  and if the sum is

above the  predetermined value then the SLP is said to be activated ie Output =1.

 

Multiple Layer Perceptron - Back Propagation Algorithm. 


 

 It is a  perceptron that teams up with  additional  perceptrons stacked up in  several layers.

Each  perceptron in the  first  Layer  sends  an  output  to  the  second  layer and then the

Second layer sends the output  to  the  third layer and finally  like  this to the  final layer.

For each  signal the  perceptron uses  different  weights. every  line  going  from a 

Perceptron  freom  one layer  to  the next  layer represents  a different  output.

The Multi Layer  Perceptron  has  another  common name  called  neural  network.

It basically  consists of  two  phases  Forward  Propagation  and  Backward  Propagation

Forward Propagation :  It adds all the inputs  multiplying  by  its  respective  wights and 

                                         generates  output.

Backward  Propagation:  It sends the  errors  backward  by  assigning  them  to  each  

                                            unit  according to  its  error.If any  error  occurs  it  shows it

                                            in  three units.

                                            1. Error  in  Input  Unit

                                            2. Error  in  Output Unit

                                            3. Error  in  Hidden  Unit.


  In  real  world  Perceptrons  work  under  the  framework  of Deep Learning  such  as  

  PyTorch, TensorFlow. These  Framework work  on  the Complete  Philosophy

  of Single  Layer  Perceptron and Multiple  layer Perceptron  and  thereby  constructs

  the  Network  of  Perceptrons  automatically.


 

 

 

No comments:

Post a Comment