Perceptron Learning Algorithm and Adaline

Perceptron learning algorithm and Adaline (adaptive linear neuron) algorithm for perceptron learning.

This octave code will:

• pick a separation line L,
• generate a random set D from a 2-dimensional RxR (or from a more general Rⁿ),
• classify D using L: above/below - makes it linearly separable by construction
• run PLA algorithm on the set D
• alternatively, run Adaline algorithm on the set D
• plot estimated separation line

How to run:

Running:

simpla(50);

will perform the above actions setting the sample size to 50, using PLA algorithm.

Running:

simAdaline(50); 

will do same as above, except it will use Adaline algorithm (adaptive linear neuron).

Running:

simAdaline(50, 0.001); 

will do same as above, except it will use multiplier of 0.001 in the Adaline algorithm.

Running:

Win = [4, -2, 0, 6, 1]';
[x,y] = generateLSsample(Win, 100); 
[Wout,r,n] = pla(x,y);

will generate 100 samples in an arbitrary dimension Rⁿ, implicitly given by vector Win, and will perform PLA algorithm on this dataset.

R-bee