You are on page 1of 2

Procedure to create, train, simulate neural network using nntool box

1) Before going to use nntool box load the training and simulating data into the workspace.

2) Type nntool on command window that will open tool box window.

3) Click on import and load the input data, target data and simulating data.

4) Click on new to create desired neural network , that will open new window.

5) Type name of the creating neural network , select the desired network type from the
available options and select the training input and target data from the loaded data.

6) Select the number of hidden layers and hidden neurons (This option is available based on
the selected network type)

7) Click on view , it will display the constructed network .

8) Click on create will create network and close the window.

9) Open the created network , train it and simulate.

10) Click on the export and save the trained network to reuse again.

Example problems:

I. And gate using perceptron neural network:

1) Run and_datagenerator.m that will load data into workspace.

2) Follow nntool box procedure.

II. Wine classification using Back propagation algorithm(BPA) neural network:

Wine classification example has 13 inputs ,3 outputs and 178 patterns . Out of which one pattern
is randomly selected to simulate and remaining are used for training.

1) Run wine_datagenerator.m that will load data into workspace.

2) Follow nntool box procedure.


III. Clustering using Radial basis function(RBF) network:

1) Run simplecluster_datagenerator.m that will load data into workspace.

2) Follow nntool box procedure.

You might also like