INT301 Lab 4-Further Practice with Perceptrons Solved

Description

Further Practice with Perceptrons

Demo:

Analyze the given program PerceptronExample, which consists of following parts:

1. (1)  Prepare mydata by generating 2 dimensional linearly separable data.

           mydata = rand(500,2);
   

           % Separate the data into two classes
   

           acceptindex = abs(mydata(:,1)-mydata(:,2))>0.012;
           mydata = mydata(acceptindex,:); % data
           myclasses = mydata(:,1)>mydata(:,2); % labels
   

          [m n]=size(mydata);
   

   You may check the data distribution using:

             scatter(mydata(:,1),mydata(:,2))
   

2. (2)  Train the perceptron by calling the function PerceptronTrn with the prepared training data (x, y), which will return the connection weights, the bias, and the number of iteration;
3. (3)  Test the trained Perceptron model with the testing data (xt, yt), by calling another function PerceptronTst, which will return the testing error;
4. (4)  Display the two classes of data points with a separating line.

Exercise:

1. (1)  The Perceptron training function uses a learning rate 0.5 and a threshold 0.5. Change these two parameters, e.g. learning rate 0.1 and threshold 0, and observe the differences;
2. (2)  Revise the program to calculate the Root-Mean-Square (RMS) error for every input data points and display the error curve, i.e. RMS vs. iteration.