apunkt/GPy
1
0
Fork 0
mirror of https://github.com/SheffieldML/GPy.git synced 2026-06-14 15:25:15 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Not needed

Browse source
This commit is contained in:
Ricardo Andrade 2013-03-27 15:06:21 +00:00
parent 67d3ba6700
commit bf9e6cbd5f
3 changed files with 0 additions and 173 deletions
Download patch file Download diff file Expand all files Collapse all files

53
GPy/examples/fitc_class_01.py
View file

@ -1,53 +0,0 @@
import pylab as pb
import numpy as np
import GPy
pb.ion()
pb.close('all')
"""
Simple 1D classification example
:param seed : seed value for data generation (default is 4).
:type seed: int
"""
seed=10000
#data = GPy.util.datasets.toy_linear_1d_classification(seed=seed)
#X = data['X']
#Y = data['Y'][:, 0:1]
#Y[Y == -1] = 0
X = np.vstack((np.random.uniform(0,10,(10,1)),np.random.uniform(7,17,(10,1)),np.random.uniform(15,25,(10,1))))
Y = np.vstack((np.zeros((10,1)),np.ones((10,1)),np.zeros((10,1))))
# Kernel object
kernel = GPy.kern.rbf(1) + GPy.kern.white(1)
# Likelihood object
distribution = GPy.likelihoods.likelihood_functions.probit()
likelihood = GPy.likelihoods.EP(Y,distribution)
Z = np.random.uniform(X.min(),X.max(),(10,1))
#Z = np.array([0,20])[:,None]
print Z
# Model definition
m = GPy.models.generalized_FITC(X,likelihood=likelihood,kernel=kernel,Z=Z,normalize_X=False)
m.set('len',2.)
m.ensure_default_constraints()
# Optimize
#m.constrain_fixed('iip')
m.update_likelihood_approximation()
print m.checkgrad(verbose=1)
# Parameters optimization:
#m.optimize()
m.pseudo_EM() #FIXME
# Plot
pb.subplot(211)
m.plot_f()
pb.subplot(212)
m.plot()
print(m)

67
GPy/examples/fitc_class_02.py
View file

@ -1,67 +0,0 @@
import pylab as pb
import numpy as np
import GPy
pb.close('all')
seed=10000
"""Run a Gaussian process classification on the crescent data. The demonstration calls the basic GP classification model and uses EP to approximate the likelihood.
:param model_type: type of model to fit ['Full', 'FITC', 'DTC'].
:param seed : seed value for data generation.
:type seed: int
:param inducing : number of inducing variables (only used for 'FITC' or 'DTC').
:type inducing: int
"""
data = GPy.util.datasets.crescent_data(seed=seed)
# Kernel object
kernel = GPy.kern.rbf(data['X'].shape[1]) + GPy.kern.white(data['X'].shape[1])
# Likelihood object
distribution = GPy.likelihoods.likelihood_functions.probit()
likelihood = GPy.likelihoods.EP(data['Y'],distribution)
sample = np.random.randint(0,data['X'].shape[0],10)
Z = data['X'][sample,:]
# create sparse GP EP model
m = GPy.models.generalized_FITC(data['X'],likelihood=likelihood,kernel=kernel,Z=Z)
m.ensure_default_constraints()
m.set('len',10.)
#m.update_likelihood_approximation()
# optimize
#m.optimize()
m.pseudo_EM()
print(m)
# plot
m.plot()
fitc = m
pb.figure()
# Kernel object
kernel = GPy.kern.rbf(data['X'].shape[1]) + GPy.kern.white(data['X'].shape[1])
# Likelihood object
distribution = GPy.likelihoods.likelihood_functions.probit()
likelihood = GPy.likelihoods.EP(data['Y'],distribution)
#sample = np.random.randint(0,data['X'].shape[0],10)
#Z = data['X'][sample,:]
# create sparse GP EP model
m = GPy.models.sparse_GP(data['X'],likelihood=likelihood,kernel=kernel,Z=Z)
m.ensure_default_constraints()
m.set('len',10.)
#m.update_likelihood_approximation()
# optimize
#m.optimize()
m.pseudo_EM()
print(m)
# plot
m.plot()
variational = m

53
GPy/examples/fitc_regr_01.py
View file

@ -1,53 +0,0 @@
import pylab as pb
import numpy as np
import GPy
pb.ion()
pb.close('all')
N = 400
M = 10
# sample inputs and outputs
X = np.random.uniform(-3.,3.,(N,1))
Y = np.sin(X)+np.random.randn(N,1)*0.05
"""Run a 1D example of a sparse GP regression."""
"""
rbf = GPy.kern.rbf(1)
noise = GPy.kern.white(1)
kernel = rbf + noise
Z = np.random.uniform(-3,3,(M,1))
likelihood = GPy.likelihoods.Gaussian(Y)
m = GPy.models.sparse_GP(X, likelihood, kernel, Z)
m.scale_factor=10000
m.constrain_positive('(variance|lengthscale|precision)')
m.checkgrad(verbose=1)
m.optimize('tnc', messages = 1)
pb.figure()
m.plot()
variational = m
"""
# construct kernel
rbf = GPy.kern.rbf(1)
noise = GPy.kern.white(1)
kernel = rbf + noise
#Z = np.random.uniform(-3,3,(M,1))
Z = variational.Z
likelihood = GPy.likelihoods.Gaussian(Y)
# create simple GP model
m = GPy.models.generalized_FITC(X, likelihood, kernel, Z=Z)
m.constrain_positive('(variance|lengthscale|precision)')
#m.constrain_fixed('iip')
m.checkgrad(verbose=1)
m.optimize('tnc', messages = 1)
#pb.figure()
#m.plot()
"""
Xnew = X.copy().flatten()
Xnew.sort()
Xnew = Xnew[:,None]
mean,var,low,up = m.predict(Xnew)
GPy.util.plot.gpplot(Xnew,mean,low,up)
fitc = m
"""