PyRadbas Home Page

Simple Radial Basis Function Networks in Python

PyRadbas is a small Python package that provides algorithms for building both exact and inexact RBFNs (radial basis function network). You can also use PyRadbas to import and run in Python any of your Matlab® RBFN.

PyRadbas is a free software licensed under the FreeBSD License. You can use and modify PyRadbas as you like... But if you need some feature, bugfix or got some suggestions, you are strongly encouraged to contribute :)


import numpy as np
import matplotlib.pyplot as plt
import pyradbas as pyrb

#Defines the mesh
x = np.linspace(-1.5, 1.5, 40)
x, y = np.meshgrid(x, x)
P = np.vstack([x.flatten(), y.flatten()])

#Defines the function
heart = lambda x,y: np.abs(x**2.+2*(y-0.5*np.sqrt(np.abs(x)))**2.-1)

#Evaluates function over every point of the grid
V = heart(P[0:1], P[1:])

plt.plot(P[0:1][V<0.2], P[1:][V<0.2], '*r', P[0:1][V>=0.2], P[1:][V>=0.2], 'o')
#defines an exact RBFN
enet = pyrb.train_exact(P.T, V.T, 0.3)

S = enet.sim(P.T).T
plt.plot(P[0:1][S<0.2], P[1:][S<0.2], '*r', P[0:1][S>=0.2], P[1:][S>=0.2], 'o') # small differences are due to ill conditioning
#What if we compute points outside training set
O=np.random.uniform(size=(2,5000), low=-2., high=2.)
S = enet.sim(O.T).T
plt.plot(O[0:1][S<0.2], O[1:][S<0.2], '*r', O[0:1][S>=0.2], O[1:][S>=0.2], 'o')
#To achieve better generalization we can train an inexact RBFN
inet = pyrb.train_ols(P.T, V.T, 0.0007, 0.3, verbose=True)

S = inet.sim(P.T).T
plt.plot(P[0:1][S<0.2], P[1:][S<0.2], '*r', P[0:1][S>=0.2], P[1:][S>=0.2], 'o')
#Outside training set...
S = inet.sim(O.T).T
plt.plot(O[0:1][S<0.2], O[1:][S<0.2], '*r', O[0:1][S>=0.2], O[1:][S>=0.2], 'o')