apunkt/GPy
1
0
Fork 0
mirror of https://github.com/SheffieldML/GPy.git synced 2026-05-10 20:42:39 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Changes to allow multiple output plotting

Browse source
This commit is contained in:
Ricardo 2013-07-31 19:00:54 +01:00
parent 7e1e8de5e4
commit 1c2a4c5c64
6 changed files with 109 additions and 27 deletions
Download patch file Download diff file Expand all files Collapse all files

1
GPy/models/__init__.py
View file

@ -12,3 +12,4 @@ from warped_gp import WarpedGP
from bayesian_gplvm import BayesianGPLVM
from mrd import MRD
from gp_multioutput import GPMultioutput
from sparse_gp_multioutput import SparseGPMultioutput

72
GPy/models/gp_multioutput.py
View file

@ -19,7 +19,7 @@ class GPMultioutput(GP):
:param X_list: input observations
:param Y_list: observed values
:param L_list: a GPy likelihood, defaults to Binomial with probit link_function
:param kernel: a GPy kernel, defaults to rbf
:param kernel_list: a GPy kernel, defaults to rbf
:param normalize_X: whether to normalize the input data before computing (predictions will be in original scales)
:type normalize_X: False|True
:param normalize_Y: whether to normalize the input data before computing (predictions will be in original scales)
@ -29,28 +29,64 @@ class GPMultioutput(GP):
"""
def __init__(self,X_list,Y_list=None,likelihood=None,kernel=None,normalize_X=False,normalize_Y=False,W=1):
def __init__(self,X_list,Y_list,noise_list=[],kernel_list=None,normalize_X=False,normalize_Y=False,W=1): #TODO W
if likelihood is None:
noise_model_list = [likelihoods.gaussian(variance=1.) for Y in Y_list]
likelihood = likelihoods.EP_Mixed_Noise(Y_list, noise_model_list)
assert len(X_list) == len(Y_list)
index = []
i = 0
for x,y in zip(X_list,Y_list):
assert x.shape[0] == y.shape[0]
index.append(np.repeat(i,y.size)[:,None])
i += 1
index = np.vstack(index)
elif Y_list is not None:
if not all(np.vstack(Y_list).flatten() == likelihood.data.flatten()):
raise Warning, 'likelihood.data and Y_list values are different.'
if noise_list == []:
likelihood_list = []
for Y in Y_list:
likelihood_list.append(likelihoods.Gaussian(Y,normalize = normalize_Y))
X = np.hstack([np.vstack(X_list),likelihood.index])
Y = np.vstack([l_.Y for l_ in likelihood_list])
likelihood = likelihoods.Gaussian(Y,normalize=False)
likelihood.index = index
if kernel is None:
X = np.hstack([np.vstack(X_list),index])
if kernel_list is None:
original_dim = X.shape[1]-1
kernel = kern.rbf(original_dim) + kern.white(original_dim)
mkernel = kernel.prod(kern.coregionalise(len(X_list),W),tensor=True) #TODO W
#kern1 = kern.rbf(1) + kern.white(1)
#kern2 = kern.coregionalise(2,1)
#kern3 = kern1.prod(kern2,tensor=True)
kernel_list = [kern.rbf(original_dim) + kern.white(original_dim)]
mkernel = kernel_list[0].prod(kern.coregionalise(len(X_list),W),tensor=True)
for k in kernel_list[1:]:
mkernel += k.prod(kern.coregionalise(len(X_list),W),tensor=True)
self.multioutput = True
GP.__init__(self, X, likelihood, mkernel, normalize_X=normalize_X)
self.ensure_default_constraints()
"""
if likelihood is None:
noise_model_list = []
for Y in Y_list:
noise_model_list.append(likelihoods.Gaussian(Y,normalize = normalize_Y))
#noise_model_list = [likelihoods.gaussian(variance=1.) for Y in Y_list]
#likelihood = likelihoods.EP_Mixed_Noise(Y_list, noise_model_list)
elif Y_list is not None:
if not all(np.vstack(Y_list).flatten() == likelihood.data.flatten()):
raise Warning, 'likelihood.data and Y_list values are different.'
X = np.hstack([np.vstack(X_list),likelihood.index])
if kernel_list is None:
original_dim = X.shape[1]-1
kernel_list = [kern.rbf(original_dim) + kern.white(original_dim)]
mkernel = kernel_list[0].prod(kern.coregionalise(len(X_list),W),tensor=True) #TODO W
for k in kernel_list[1:]:
mkernel += k.prod(kern.coregionalise(len(X_list),W),tensor=True) #TODO W
#kern1 = kern.rbf(1) + kern.white(1)
#kern2 = kern.coregionalise(2,1)
#kern3 = kern1.prod(kern2,tensor=True)
"""