apunkt/GPy
1
0
Fork 0
mirror of https://github.com/SheffieldML/GPy.git synced 2026-04-28 14:26:23 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Fixed some examples and tests, and stated that Y metadata doesnt need to be the same size as Y

Browse source
This commit is contained in:
Alan Saul 2014-02-12 18:02:24 +00:00
parent 646400f49e
commit d2e8807a88
3 changed files with 21 additions and 24 deletions
Download patch file Download diff file Expand all files Collapse all files

1
GPy/core/gp.py
View file

@ -39,7 +39,6 @@ class GP(Model):
_, self.output_dim = self.Y.shape _, self.output_dim = self.Y.shape
if Y_metadata is not None: if Y_metadata is not None:
assert Y_metadata.shape == self.Y.shape
self.Y_metadata = ObservableArray(Y_metadata) self.Y_metadata = ObservableArray(Y_metadata)
else: else:
self.Y_metadata = None self.Y_metadata = None

40
GPy/examples/non_gaussian.py
View file

@ -42,38 +42,35 @@ def student_t_approx(optimize=True, plot=True):
kernel4 = GPy.kern.rbf(X.shape[1]) + GPy.kern.white(X.shape[1]) kernel4 = GPy.kern.rbf(X.shape[1]) + GPy.kern.white(X.shape[1])
#Gaussian GP model on clean data #Gaussian GP model on clean data
m1 = GPy.models.GPRegression(X, Y.copy(), kernel=kernel1) #m1 = GPy.models.GPRegression(X, Y.copy(), kernel=kernel1)
# optimize ## optimize
m1.ensure_default_constraints() #m1['white'].constrain_fixed(1e-5)
m1['white'] = 1e-5 #m1.randomize()
m1['white'].constrain_fixed('white')
m1.randomize()
#Gaussian GP model on corrupt data ##Gaussian GP model on corrupt data
m2 = GPy.models.GPRegression(X, Yc.copy(), kernel=kernel2) #m2 = GPy.models.GPRegression(X, Yc.copy(), kernel=kernel2)
m2.ensure_default_constraints() #m1['white'].constrain_fixed(1e-5)
m1['white'] = 1e-5 #m2.randomize()
m1['white'].constrain_fixed('white')
m2.randomize()
#Student t GP model on clean data #Student t GP model on clean data
t_distribution = GPy.likelihoods.StudentT(deg_free=deg_free, sigma2=edited_real_sd) t_distribution = GPy.likelihoods.StudentT(deg_free=deg_free, sigma2=edited_real_sd)
laplace_inf = GPy.inference.latent_function_inference.LaplaceInference() laplace_inf = GPy.inference.latent_function_inference.Laplace()
m3 = GPy.core.GP(X, Y.copy(), kernel3, likelihood=t_distribution, inference_method=laplace_inf) m3 = GPy.core.GP(X, Y.copy(), kernel3, likelihood=t_distribution, inference_method=laplace_inf)
m3.ensure_default_constraints()
m3['t_noise'].constrain_bounded(1e-6, 10.) m3['t_noise'].constrain_bounded(1e-6, 10.)
m3['white'] = 1e-5 m3['white'].constrain_fixed(1e-5)
m3['white'].constrain_fixed()
m3.randomize() m3.randomize()
debug = True
print m3
if debug:
m3.optimize(messages=1)
return m3
#Student t GP model on corrupt data #Student t GP model on corrupt data
t_distribution = GPy.likelihoods.StudentT(deg_free=deg_free, sigma2=edited_real_sd) t_distribution = GPy.likelihoods.StudentT(deg_free=deg_free, sigma2=edited_real_sd)
laplace_inf = GPy.inference.latent_function_inference.LaplaceInference() laplace_inf = GPy.inference.latent_function_inference.Laplace()
m4 = GPy.core.GP(X, Yc.copy(), kernel4, likelihood=t_distribution, inference_method=laplace_inf) m4 = GPy.core.GP(X, Yc.copy(), kernel4, likelihood=t_distribution, inference_method=laplace_inf)
m4.ensure_default_constraints()
m4['t_noise'].constrain_bounded(1e-6, 10.) m4['t_noise'].constrain_bounded(1e-6, 10.)
m4['white'] = 1e-5 m4['white'].constrain_fixed(1e-5)
m4['white'].constrain_fixed()
m4.randomize() m4.randomize()
if optimize: if optimize:
@ -156,7 +153,6 @@ def boston_example(optimize=True, plot=True):
#Gaussian GP #Gaussian GP
print "Gauss GP" print "Gauss GP"
mgp = GPy.models.GPRegression(X_train.copy(), Y_train.copy(), kernel=kernelgp.copy()) mgp = GPy.models.GPRegression(X_train.copy(), Y_train.copy(), kernel=kernelgp.copy())
mgp.ensure_default_constraints()
mgp.constrain_fixed('white', 1e-5) mgp.constrain_fixed('white', 1e-5)
mgp['rbf_len'] = rbf_len mgp['rbf_len'] = rbf_len
mgp['noise'] = noise mgp['noise'] = noise
@ -174,7 +170,6 @@ def boston_example(optimize=True, plot=True):
g_distribution = GPy.likelihoods.noise_model_constructors.gaussian(variance=noise, N=N, D=D) g_distribution = GPy.likelihoods.noise_model_constructors.gaussian(variance=noise, N=N, D=D)
g_likelihood = GPy.likelihoods.Laplace(Y_train.copy(), g_distribution) g_likelihood = GPy.likelihoods.Laplace(Y_train.copy(), g_distribution)
mg = GPy.models.GPRegression(X_train.copy(), Y_train.copy(), kernel=kernelstu.copy(), likelihood=g_likelihood) mg = GPy.models.GPRegression(X_train.copy(), Y_train.copy(), kernel=kernelstu.copy(), likelihood=g_likelihood)
mg.ensure_default_constraints()
mg.constrain_positive('noise_variance') mg.constrain_positive('noise_variance')
mg.constrain_fixed('white', 1e-5) mg.constrain_fixed('white', 1e-5)
mg['rbf_len'] = rbf_len mg['rbf_len'] = rbf_len
@ -194,7 +189,6 @@ def boston_example(optimize=True, plot=True):
t_distribution = GPy.likelihoods.noise_model_constructors.student_t(deg_free=df, sigma2=noise) t_distribution = GPy.likelihoods.noise_model_constructors.student_t(deg_free=df, sigma2=noise)
stu_t_likelihood = GPy.likelihoods.Laplace(Y_train.copy(), t_distribution) stu_t_likelihood = GPy.likelihoods.Laplace(Y_train.copy(), t_distribution)
mstu_t = GPy.models.GPRegression(X_train.copy(), Y_train.copy(), kernel=kernelstu.copy(), likelihood=stu_t_likelihood) mstu_t = GPy.models.GPRegression(X_train.copy(), Y_train.copy(), kernel=kernelstu.copy(), likelihood=stu_t_likelihood)
mstu_t.ensure_default_constraints()
mstu_t.constrain_fixed('white', 1e-5) mstu_t.constrain_fixed('white', 1e-5)
mstu_t.constrain_bounded('t_noise', 0.0001, 1000) mstu_t.constrain_bounded('t_noise', 0.0001, 1000)
mstu_t['rbf_len'] = rbf_len mstu_t['rbf_len'] = rbf_len

4
GPy/testing/likelihood_tests.py
View file

@ -540,6 +540,10 @@ class TestNoiseModels(object):
#import ipdb; ipdb.set_trace() #import ipdb; ipdb.set_trace()
#NOTE this test appears to be stochastic for some likelihoods (student t?) #NOTE this test appears to be stochastic for some likelihoods (student t?)
# appears to all be working in test mode right now... # appears to all be working in test mode right now...
if not m.checkgrad():
import ipdb; ipdb.set_trace() # XXX BREAKPOINT
assert m.checkgrad(step=step) assert m.checkgrad(step=step)
########### ###########