Scribd
Upload
62 views18 pages

ATC-Alat Berat Part 4

The document discusses using MATLAB to program neural networks for classification problems, including creating single neurons, custom networks, perceptrons for linearly separable and 4-class data, and multilayer perceptrons to solve XOR and 4-class problems. It provides code examples for setting up input and output data, defining and training different neural network architectures, and evaluating their performance on classification tasks. The training works by adjusting the network weights and biases to minimize the error between the network's predictions and the actual target outputs for the training data.

Uploaded by

KABINET JALADARA NABDA
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
62 views18 pages

ATC-Alat Berat Part 4

The document discusses using MATLAB to program neural networks for classification problems, including creating single neurons, custom networks, perceptrons for linearly separable and 4-class data, and multilayer perceptrons to solve XOR and 4-class problems. It provides code examples for setting up input and output data, defining and training different neural network architectures, and evaluating their performance on classification tasks. The training works by adjusting the network weights and biases to minimize the error between the network's predictions and the actual target outputs for the training data.

Uploaded by

KABINET JALADARA NABDA
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 18

Aplikasi Teknologi Cerdas

Neural Network Practical : MATLAB programming


1. Neuron Output

FIGURE 1 A SINGLE-INPUT NEURON

p : Input vector
a : Output vector
w : Weight
b : Bias
f : Activation Function/ Transfer Function

Problem Description: Calculate the output of a simple neuron

MATLAB Code :

Result :
2. Create a Neural Network

Problem Description: Create and view custom Neural Networks

 Define one sample: inputs and outputs

 Define and custom network

You will get:


 Define Topology and Transfer Function/ Activation Function

 Configure Network

 Train net and Calculate Neuron Output

Train function : “Levenberg-Marquardt”

Performance Function : “Mean Squared Error”


3. Classification of Linearly Separable Data with A Perceptron

Problem Description: Two clusters of data, belonging to two classes, are defined in a 2-dimensional input space. Classes
are linearly separable. The task is to construct a Perceptron for the classification of data

 Define Input and Output Data

You will get this plot:

 Create and Train Perceptron


 Plot Decision Boundary

You will get:


4. Classification of 4-class Problem with A Perceptron

Problem Description: Perceptron network with 2-inputs and 2-outputs is trained to classify input vectors into 4 categories

 Define Data
 Prepare inputs and outputs for perceptron training

 Create a perceptron

 Train a perceptron

Result:
 How to use trained perceptron

This p is categorized to Class B because its output y = [1 1]T, you can confirm it by looking for p’s
coordinate in the result plot.
5. Solving XOR Problem with a Multilayer Perceptron

Problem Description: 4 clusters of data (A,B,C,D) are defined in a 2-dimensional input space. (A,C) and (B,D) clusters
represent XOR classification problem. The task is to define a neural network for solving the XOR problem.

 Define 4 clusters of input data

 Define output coding for XOR problem


 Prepare inputs and outputs for network training

 Create and train a multilayer perceptron

 Plot targets and network response to see how good the network learns the data
 Plot Classification result for the complete input space
6. Classification of a 4-Class Problem with a Multilayer
Perceptron

Problem Description: 4 clusters of data (A,B,C,D) are defined in a 2-dimensional input space. The task is to define a neural
network for classification of arbitrary point in the 2-dimensional space into one of the classes (A,B,C,D)

 Define 4 Clusters of input data

 Define output coding for all 4 clusters


 Prepare inputs and ouputs for network training

 Create and train a multilayer perceptron

 Evaluate Network performance and plot results


 Plot classification result for the complete input space
Tasks:

1. Try all of the problem above


2. Try to solve another problem with at least 3 various inputs (2-dimensions). One of them should be based on 2 last
digits of your NRP. Solve It with MATLAB code.
(For Example: NRP 10311710000487 use input p = [8 7])
3. Compare all of your results in MATLAB with your manual calculations.
4. How does the training work in the neural network?

You might also like