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

more coopyrighting

Browse source
This commit is contained in:
James Hensman 2014-11-21 11:52:28 +00:00
parent 6aae3a37c8
commit 384b6c70c5
8 changed files with 30 additions and 21 deletions
Download patch file Download diff file Expand all files Collapse all files

3
GPy/inference/latent_function_inference/__init__.py
View file

@ -1,3 +1,6 @@
# Copyright (c) 2012, James Hensman
# Licensed under the BSD 3-clause license (see LICENSE.txt)
__doc__ = """ __doc__ = """
Inference over Gaussian process latent functions Inference over Gaussian process latent functions

2
GPy/inference/latent_function_inference/dtc.py
View file

@ -1,4 +1,4 @@
# Copyright (c) 2012, James Hensman # Copyright (c) 2012-2014, James Hensman
# Licensed under the BSD 3-clause license (see LICENSE.txt) # Licensed under the BSD 3-clause license (see LICENSE.txt)
from posterior import Posterior from posterior import Posterior

2
GPy/inference/latent_function_inference/exact_gaussian_inference.py
View file

@ -1,4 +1,4 @@
# Copyright (c) 2012, GPy authors (see AUTHORS.txt). # Copyright (c) 2012-2014, GPy authors (see AUTHORS.txt).
# Licensed under the BSD 3-clause license (see LICENSE.txt) # Licensed under the BSD 3-clause license (see LICENSE.txt)
from posterior import Posterior from posterior import Posterior

2
GPy/inference/latent_function_inference/expectation_propagation.py
View file

@ -1,3 +1,5 @@
# Copyright (c) 2012-2014, GPy authors (see AUTHORS.txt).
# Licensed under the BSD 3-clause license (see LICENSE.txt)
import numpy as np import numpy as np
from ...util.linalg import pdinv,jitchol,DSYR,tdot,dtrtrs, dpotrs from ...util.linalg import pdinv,jitchol,DSYR,tdot,dtrtrs, dpotrs
from posterior import Posterior from posterior import Posterior

3
GPy/inference/latent_function_inference/expectation_propagation_dtc.py
View file

@ -1,3 +1,6 @@
# Copyright (c) 2012-2014, GPy authors (see AUTHORS.txt).
# Licensed under the BSD 3-clause license (see LICENSE.txt)
import numpy as np import numpy as np
from ...util import diag from ...util import diag
from ...util.linalg import mdot, jitchol, backsub_both_sides, tdot, dtrtrs, dtrtri, dpotri, dpotrs, symmetrify, DSYR from ...util.linalg import mdot, jitchol, backsub_both_sides, tdot, dtrtrs, dtrtri, dpotri, dpotrs, symmetrify, DSYR

35
GPy/inference/latent_function_inference/inferenceX.py
View file

@ -1,5 +1,6 @@
""" # Copyright (c) 2014, Zhenwen Dai
""" # Licensed under the BSD 3-clause license (see LICENSE.txt)
import numpy as np import numpy as np
from ...core import Model from ...core import Model
from ...core.parameterization import variational from ...core.parameterization import variational
@ -7,27 +8,27 @@ from ...core.parameterization import variational
def infer_newX(model, Y_new, optimize=True, init='L2'): def infer_newX(model, Y_new, optimize=True, init='L2'):
""" """
Infer the distribution of X for the new observed data *Y_new*. Infer the distribution of X for the new observed data *Y_new*.
:param model: the GPy model used in inference :param model: the GPy model used in inference
:type model: GPy.core.Model :type model: GPy.core.Model
:param Y_new: the new observed data for inference :param Y_new: the new observed data for inference
:type Y_new: numpy.ndarray :type Y_new: numpy.ndarray
:param optimize: whether to optimize the location of new X (True by default) :param optimize: whether to optimize the location of new X (True by default)
:type optimize: boolean :type optimize: boolean
:return: a tuple containing the estimated posterior distribution of X and the model that optimize X :return: a tuple containing the estimated posterior distribution of X and the model that optimize X
:rtype: (GPy.core.parameterization.variational.VariationalPosterior, GPy.core.Model) :rtype: (GPy.core.parameterization.variational.VariationalPosterior, GPy.core.Model)
""" """
infr_m = InferenceX(model, Y_new, init=init) infr_m = InferenceX(model, Y_new, init=init)
if optimize: if optimize:
infr_m.optimize() infr_m.optimize()
return infr_m.X, infr_m return infr_m.X, infr_m
class InferenceX(Model): class InferenceX(Model):
""" """
The class for inference of new X with given new Y. (do_test_latent) The class for inference of new X with given new Y. (do_test_latent)
:param model: the GPy model used in inference :param model: the GPy model used in inference
:type model: GPy.core.Model :type model: GPy.core.Model
:param Y: the new observed data for inference :param Y: the new observed data for inference
@ -67,12 +68,12 @@ class InferenceX(Model):
self.Y = Y self.Y = Y
self.X = self._init_X(model, Y, init=init) self.X = self._init_X(model, Y, init=init)
self.compute_dL() self.compute_dL()
self.link_parameter(self.X) self.link_parameter(self.X)
def _init_X(self, model, Y_new, init='L2'): def _init_X(self, model, Y_new, init='L2'):
# Initialize the new X by finding the nearest point in Y space. # Initialize the new X by finding the nearest point in Y space.
Y = model.Y Y = model.Y
if self.missing_data: if self.missing_data:
Y = Y[:,self.valid_dim] Y = Y[:,self.valid_dim]
@ -86,7 +87,7 @@ class InferenceX(Model):
elif init=='rand': elif init=='rand':
dist = np.random.rand(Y_new.shape[0],Y.shape[0]) dist = np.random.rand(Y_new.shape[0],Y.shape[0])
idx = dist.argmin(axis=1) idx = dist.argmin(axis=1)
from ...models import SSGPLVM from ...models import SSGPLVM
from ...util.misc import param_to_array from ...util.misc import param_to_array
if isinstance(model, SSGPLVM): if isinstance(model, SSGPLVM):
@ -99,9 +100,9 @@ class InferenceX(Model):
else: else:
from ...core import Param from ...core import Param
X = Param('latent mean',param_to_array(model.X[idx]).copy()) X = Param('latent mean',param_to_array(model.X[idx]).copy())
return X return X
def compute_dL(self): def compute_dL(self):
# Common computation # Common computation
beta = 1./np.fmax(self.likelihood.variance, 1e-6) beta = 1./np.fmax(self.likelihood.variance, 1e-6)
@ -120,7 +121,7 @@ class InferenceX(Model):
self.dL_dpsi2 = beta*(output_dim*self.posterior.woodbury_inv - np.einsum('md,od->mo',wv, wv))/2. self.dL_dpsi2 = beta*(output_dim*self.posterior.woodbury_inv - np.einsum('md,od->mo',wv, wv))/2.
self.dL_dpsi1 = beta*np.dot(self.Y, wv.T) self.dL_dpsi1 = beta*np.dot(self.Y, wv.T)
self.dL_dpsi0 = -beta/2.*output_dim* np.ones(self.Y.shape[0]) self.dL_dpsi0 = -beta/2.*output_dim* np.ones(self.Y.shape[0])
def parameters_changed(self): def parameters_changed(self):
if self.uncertain_input: if self.uncertain_input:
psi0 = self.kern.psi0(self.Z, self.X) psi0 = self.kern.psi0(self.Z, self.X)
@ -132,7 +133,7 @@ class InferenceX(Model):
psi2 = np.dot(psi1.T,psi1) psi2 = np.dot(psi1.T,psi1)
self._log_marginal_likelihood = (self.dL_dpsi2*psi2).sum()+(self.dL_dpsi1*psi1).sum()+(self.dL_dpsi0*psi0).sum() self._log_marginal_likelihood = (self.dL_dpsi2*psi2).sum()+(self.dL_dpsi1*psi1).sum()+(self.dL_dpsi0*psi0).sum()
if self.uncertain_input: if self.uncertain_input:
X_grad = self.kern.gradients_qX_expectations(variational_posterior=self.X, Z=self.Z, dL_dpsi0=self.dL_dpsi0, dL_dpsi1=self.dL_dpsi1, dL_dpsi2=self.dL_dpsi2) X_grad = self.kern.gradients_qX_expectations(variational_posterior=self.X, Z=self.Z, dL_dpsi0=self.dL_dpsi0, dL_dpsi1=self.dL_dpsi1, dL_dpsi2=self.dL_dpsi2)
self.X.set_gradients(X_grad) self.X.set_gradients(X_grad)
@ -141,7 +142,7 @@ class InferenceX(Model):
X_grad = self.kern.gradients_X_diag(self.dL_dpsi0, self.X) X_grad = self.kern.gradients_X_diag(self.dL_dpsi0, self.X)
X_grad += self.kern.gradients_X(dL_dpsi1, self.X, self.Z) X_grad += self.kern.gradients_X(dL_dpsi1, self.X, self.Z)
self.X.gradient = X_grad self.X.gradient = X_grad
if self.uncertain_input: if self.uncertain_input:
from ...core.parameterization.variational import SpikeAndSlabPrior from ...core.parameterization.variational import SpikeAndSlabPrior
if isinstance(self.variational_prior, SpikeAndSlabPrior): if isinstance(self.variational_prior, SpikeAndSlabPrior):
@ -155,7 +156,7 @@ class InferenceX(Model):
# update for the KL divergence # update for the KL divergence
self.variational_prior.update_gradients_KL(self.X) self.variational_prior.update_gradients_KL(self.X)
self._log_marginal_likelihood += -KL_div self._log_marginal_likelihood += -KL_div
def log_likelihood(self): def log_likelihood(self):
return self._log_marginal_likelihood return self._log_marginal_likelihood

2
GPy/inference/latent_function_inference/laplace.py
View file

@ -1,4 +1,4 @@
# Copyright (c) 2013, 2014 GPy authors (see AUTHORS.txt). # Copyright (c) 2013, 2014 Alan Saul
# Licensed under the BSD 3-clause license (see LICENSE.txt) # Licensed under the BSD 3-clause license (see LICENSE.txt)
# #
#Parts of this file were influenced by the Matlab GPML framework written by #Parts of this file were influenced by the Matlab GPML framework written by

2
GPy/inference/latent_function_inference/var_dtc_parallel.py
View file

@ -1,4 +1,4 @@
# Copyright (c) 2012, GPy authors (see AUTHORS.txt). # Copyright (c) 2014, GPy authors (see AUTHORS.txt).
# Licensed under the BSD 3-clause license (see LICENSE.txt) # Licensed under the BSD 3-clause license (see LICENSE.txt)
from posterior import Posterior from posterior import Posterior