apunkt/GPy
1
0
Fork 0
mirror of https://github.com/SheffieldML/GPy.git synced 2026-04-29 14:56:24 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Just breaking some things...

Browse source
This commit is contained in:
Alan Saul 2013-03-19 11:47:53 +00:00
parent 2bf1cf0eb6
commit 46d59c94b2
3 changed files with 113 additions and 43 deletions
Download patch file Download diff file Expand all files Collapse all files

88
python/examples/laplace_approximations.py
View file

@ -16,47 +16,75 @@ def student_t_approx():
#Add student t random noise to datapoints
deg_free = 2.5
t_rv = t(deg_free, loc=5, scale=1)
t_rv = t(deg_free, loc=0, scale=1)
noise = t_rv.rvs(size=Y.shape)
Y += noise
#Add some extreme value noise to some of the datapoints
#percent_corrupted = 0.05
#corrupted_datums = int(np.round(Y.shape[0] * percent_corrupted))
#indices = np.arange(Y.shape[0])
#np.random.shuffle(indices)
#corrupted_indices = indices[:corrupted_datums]
#print corrupted_indices
#noise = t_rv.rvs(size=(len(corrupted_indices), 1))
#Y[corrupted_indices] += noise
# Kernel object
print X.shape
kernel = GPy.kern.rbf(X.shape[1])
#print X.shape
#kernel = GPy.kern.rbf(X.shape[1])
#A GP should completely break down due to the points as they get a lot of weight
# create simple GP model
m = GPy.models.GP_regression(X, Y, kernel=kernel)
##A GP should completely break down due to the points as they get a lot of weight
## create simple GP model
#m = GPy.models.GP_regression(X, Y, kernel=kernel)
# optimize
m.ensure_default_constraints()
m.optimize()
# plot
#m.plot()
print m
## optimize
#m.ensure_default_constraints()
#m.optimize()
## plot
##m.plot()
#print m
#with a student t distribution, since it has heavy tails it should work well
likelihood_function = student_t(deg_free, sigma=1)
lap = Laplace(Y, likelihood_function)
cov = kernel.K(X)
lap.fit_full(cov)
#Get one sample (just look at a single Y
#mode = float(lap.f_hat[0])
#variance = float((deg_free/(deg_free-2))) #BUG: Not convinced this is giving reasonable variables
#variance = float((deg_free/(deg_free-2)) + np.diagonal(lap.hess_hat)[0]) #BUG: Not convinced this is giving reasonable variables
#likelihood_function = student_t(deg_free, sigma=1)
#lap = Laplace(Y, likelihood_function)
#cov = kernel.K(X)
#lap.fit_full(cov)
test_range = np.arange(0, 10, 0.1)
plt.plot(test_range, t_rv.pdf(test_range))
for i in xrange(X.shape[0]):
mode = lap.f_hat[i]
covariance = lap.hess_hat_i[i,i]
scaling = np.exp(lap.ln_z_hat)
normalised_approx = norm(loc=mode, scale=covariance)
print "Normal with mode %f, and variance %f" % (mode, covariance)
plt.plot(test_range, normalised_approx.pdf(test_range))
plt.show()
#test_range = np.arange(0, 10, 0.1)
#plt.plot(test_range, t_rv.pdf(test_range))
#for i in xrange(X.shape[0]):
#mode = lap.f_hat[i]
#covariance = lap.hess_hat_i[i,i]
#scaling = np.exp(lap.ln_z_hat)
#normalised_approx = norm(loc=mode, scale=covariance)
#print "Normal with mode %f, and variance %f" % (mode, covariance)
#plt.plot(test_range, scaling*normalised_approx.pdf(test_range))
#plt.show()
#import ipdb; ipdb.set_trace() ### XXX BREAKPOINT
# Likelihood object
t_distribution = student_t(deg_free, sigma=1)
stu_t_likelihood = Laplace(Y, t_distribution)
kernel = GPy.kern.rbf(X.shape[1])
m = GPy.models.GP(X, stu_t_likelihood, kernel)
m.ensure_default_constraints()
m.update_likelihood_approximation()
print "NEW MODEL"
print(m)
# optimize
#m.optimize()
print(m)
# plot
m.plot()
import ipdb; ipdb.set_trace() ### XXX BREAKPOINT
return m
def noisy_laplace_approx():
"""