Home

Backpropagation Neural Network

Description

A neural network code fully writen in python

Instalation

Just download the file "Neural-Network.py" and add it into your project folder
In this case there are two examples to show how to use it

How to use

There are five simple steps to use it

• Instantiate an object
• Assign data
• Set model
• Train the network
• Predict

Example 1

Document "exampleHeightAndWeight.py"

• In this example the neural network will predict if one person is under, normal or over weight
  
• For input let's given height (cm) and weight (kg)
  
• For output let's given "1 0 0" for under weight, "0 1 0" for normal weight and "0 0 1" for over weight
  
• Note: Remember use real data to get better results.
  

Data input

Data output

Instantiate an object

Assign data

Set model

• In this case there are 8 input layers, 10 hidden and 3 of output.
• Each layer is separated with a comma.
• For input and hidden layers doesn't matter how much you wan to add.
• The size of output layer must be equals to size of row data output, we want "1 0 0", "0 1 0" or "0 0 1", so the size is 3.
  

Train the network

• Limit of iterations for the neural network
  

Predict

• Receive two values, an array with the data we want predict and a boolean.
• Return an array with solution.
• If boolean is "True" will show a message "Prediction: [values]".
• If boolean is not given or is "False" the message will not be displayed.
  

Outputs

• You can use any BIM calculator to check results.