The idea of this project is to classify 3 groups of 2-dimension data (x,y) linearly separables with a multi-output perceptron.
Build a neural network capable to classify 3 groups of data.
Fig.1 Three groups of 2 dimensional (x,y) points. Source: Samarasinghe S.[1]. Graphics by my own.
A 2-3 neural network was built. After 20 epochs and a learning rate of 0.1 the equations of the three boundaries are:
f1(x) = 3.45x -1.33
f2(x) = -1.56x + 2.59
f3(x) = 0.16x + 0.1
Fig.2 The classification boundaries superimposed on the data. Source: own elaboration.
[1] Samarasinghe, S. Neural Networks for Applied Sciences and Engineering. From Fundamentals to Complex Pattern Recognition, Auerbach Publications, pp.40-44, 2007.