apunkt/GPy
1
0
Fork 0
mirror of https://github.com/SheffieldML/GPy.git synced 2026-05-30 14:35:15 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Modified likelihoods test to better test multioutput likelihood

Browse source
This commit is contained in:
Eero Siivola 2018-06-24 12:41:36 +03:00
parent 292e69ee0d
commit 8e69db51a2
1 changed files with 13 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

17
GPy/testing/likelihood_tests.py
View file

@ -129,6 +129,15 @@ class TestNoiseModels(object):
self.Y_metadata = dict() self.Y_metadata = dict()
self.Y_metadata['censored'] = censored self.Y_metadata['censored'] = censored
self.Y_metadata['output_index'] = np.zeros((self.N,1), dtype=int) self.Y_metadata['output_index'] = np.zeros((self.N,1), dtype=int)
self.Y_metadata2 = dict()
self.Y_metadata2['censored'] = censored
inds = np.zeros((self.N,1), dtype=int)
inds[5:10] = 1
inds[10:] = 2
self.Y_metadata2['output_index'] = inds
self.combY = self.Y
self.combY[10:] = np.where(self.binary_Y[10:] >0, self.binary_Y[10:], 0)
print(self.combY)
#Make a bigger step as lower bound can be quite curved #Make a bigger step as lower bound can be quite curved
self.step = 1e-4 self.step = 1e-4
@ -294,11 +303,11 @@ class TestNoiseModels(object):
"laplace": True "laplace": True
}, },
"multioutput_default": { "multioutput_default": {
"model": GPy.likelihoods.MultioutputLikelihood([GPy.likelihoods.Bernoulli()]), "model": GPy.likelihoods.MultioutputLikelihood([GPy.likelihoods.Gaussian(), GPy.likelihoods.Poisson(), GPy.likelihoods.Bernoulli()]),
"link_f_constraints": [partial(self.constrain_bounded, lower=0, upper=1)], "link_f_constraints": [partial(self.constrain_bounded, lower=0, upper=1)],
"laplace": True, "laplace": True,
"Y": self.binary_Y, "Y": self.combY,
"Y_metadata": self.Y_metadata, "Y_metadata": self.Y_metadata2,
"ep": True, "ep": True,
"variational_expectations": True, "variational_expectations": True,
} }
@ -627,7 +636,7 @@ class TestNoiseModels(object):
# Y = Y/Y.max() # Y = Y/Y.max()
white_var = 1e-4 white_var = 1e-4
kernel = GPy.kern.RBF(X.shape[1]) + GPy.kern.White(X.shape[1]) kernel = GPy.kern.RBF(X.shape[1]) + GPy.kern.White(X.shape[1])
ep_inf = GPy.inference.latent_function_inference.EP() ep_inf = GPy.inference.latent_function_inference.EP(always_reset=True)
m = GPy.core.GP(X.copy(), Y.copy(), kernel=kernel, likelihood=model, Y_metadata=Y_metadata, inference_method=ep_inf) m = GPy.core.GP(X.copy(), Y.copy(), kernel=kernel, likelihood=model, Y_metadata=Y_metadata, inference_method=ep_inf)
m['.*white'].constrain_fixed(white_var) m['.*white'].constrain_fixed(white_var)