apunkt/GPy
1
0
Fork 0
mirror of https://github.com/SheffieldML/GPy.git synced 2026-05-10 12:32:40 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

[GPU] GPU kernel

Browse source
This commit is contained in:
Zhenwen Dai 2014-03-26 10:47:33 +00:00
parent b5b17b9715
commit 53627ee282
7 changed files with 90 additions and 60 deletions
Download patch file Download diff file Expand all files Collapse all files

50
GPy/inference/latent_function_inference/var_dtc_parallel.py
View file

@ -279,4 +279,54 @@ class VarDTC_minibatch(object):
'dL_dthetaL':dL_dthetaL}
return isEnd, (n_start,n_end), grad_dict
def update_gradients(model):
model._log_marginal_likelihood, dL_dKmm, model.posterior = model.inference_method.inference_likelihood(model.kern, model.X, model.Z, model.likelihood, model.Y)
het_noise = model.likelihood.variance.size > 1
if het_noise:
dL_dthetaL = np.empty((model.Y.shape[0],))
else:
dL_dthetaL = 0
#gradients w.r.t. kernel
model.kern.update_gradients_full(dL_dKmm, model.Z, None)
kern_grad = model.kern.gradient.copy()
#gradients w.r.t. Z
model.Z.gradient[:,model.kern.active_dims] = model.kern.gradients_X(dL_dKmm, model.Z)
isEnd = False
while not isEnd:
isEnd, n_range, grad_dict = model.inference_method.inference_minibatch(model.kern, model.X, model.Z, model.likelihood, model.Y)
if isinstance(model.X, VariationalPosterior):
#gradients w.r.t. kernel
model.kern.update_gradients_expectations(variational_posterior=model.X[n_range[0]:n_range[1]], Z=model.Z, dL_dpsi0=grad_dict['dL_dpsi0'], dL_dpsi1=grad_dict['dL_dpsi1'], dL_dpsi2=grad_dict['dL_dpsi2'])
kern_grad += model.kern.gradient
#gradients w.r.t. Z
model.Z.gradient[:,model.kern.active_dims] += model.kern.gradients_Z_expectations(
grad_dict['dL_dpsi1'], grad_dict['dL_dpsi2'], Z=model.Z, variational_posterior=model.X[n_range[0]:n_range[1]])
#gradients w.r.t. posterior parameters of X
X_grad = model.kern.gradients_qX_expectations(variational_posterior=model.X[n_range[0]:n_range[1]], Z=model.Z, dL_dpsi0=grad_dict['dL_dpsi0'], dL_dpsi1=grad_dict['dL_dpsi1'], dL_dpsi2=grad_dict['dL_dpsi2'])
model.set_X_gradients(model.X[n_range[0]:n_range[1]], X_grad)
if het_noise:
dL_dthetaL[n_range[0]:n_range[1]] = grad_dict['dL_dthetaL']
else:
dL_dthetaL += grad_dict['dL_dthetaL']
# Set the gradients w.r.t. kernel
model.kern.gradient = kern_grad
# Update Log-likelihood
model._log_marginal_likelihood -= model.variational_prior.KL_divergence(model.X)
# update for the KL divergence
model.variational_prior.update_gradients_KL(model.X)
# dL_dthetaL
model.likelihood.update_gradients(dL_dthetaL)