GPy tutorial
Basics
Very basics of GPy, interacting with models, ipynb
Regression tutorial
GP regression by Nicolas Durrande 2013, ipynb
import numpy as np
from matplotlib import pyplot as plt
import GPy
X = np.random.uniform(-3.,3.,(20,1))
# add noise into Y
Y = np.sin(X) + np.random.randn(20,1)*0.05
One trick when working with PyCharm is to disable interactive functionality in matplotlib as
plt.interactive(False)
Define the covariance kernel, i.e. RBF, and then form the GPy model m.
kernel = GPy.kern.RBF(input_dim=1, variance = 1., lengthscale= 1.)
m = GPy.models.GPRegression(X, Y, kernel)
After initialization, we can optimize
# the normal way
# m.optimize(messages=True)
# with restarts to get better results
m.optimize_restarts(num_restarts = 20)
Kernel
Kernel tutorial by Nicolas Durrande, 2013, ipynb
Combine kernels
k1 = GPy.kern.RBF(1, 1., 2.)
k2 = GPy.kern.Matern32(1, 0.5, 0.2)
# product of kernels
k_prod = k1 * k2
k_prod.plot()
# Sum of kernels
k_add = k1 + k2
k_add.plot()
The kernels that have been added are pythonic in that the objects remain linked: changing parameters of an add kernel changes those of the constituent parts, and vice versa
k_prod.rbf.variance = 12.
print k1