mirror of
https://github.com/SheffieldML/GPy.git
synced 2026-05-24 14:15:14 +02:00
Merge branch 'devel' of https://github.com/SheffieldML/GPy into devel
Conflicts: GPy/models/GPLVM.py
This commit is contained in:
Neil Lawrence
commit
b7bf712f48
35 changed files with 723 additions and 657 deletions
|
|
@ -21,13 +21,15 @@ def crescent_data(seed=default_seed): # FIXME
|
|||
"""
|
||||
|
||||
data = GPy.util.datasets.crescent_data(seed=seed)
|
||||
Y = data['Y']
|
||||
Y[Y.flatten()==-1] = 0
|
||||
|
||||
# Kernel object
|
||||
kernel = GPy.kern.rbf(data['X'].shape[1])
|
||||
|
||||
# Likelihood object
|
||||
distribution = GPy.likelihoods.likelihood_functions.probit()
|
||||
likelihood = GPy.likelihoods.EP(data['Y'], distribution)
|
||||
distribution = GPy.likelihoods.likelihood_functions.binomial()
|
||||
likelihood = GPy.likelihoods.EP(Y, distribution)
|
||||
|
||||
|
||||
m = GPy.models.GP(data['X'], likelihood, kernel)
|
||||
|
|
@ -49,12 +51,15 @@ def oil():
|
|||
Run a Gaussian process classification on the oil data. The demonstration calls the basic GP classification model and uses EP to approximate the likelihood.
|
||||
"""
|
||||
data = GPy.util.datasets.oil()
|
||||
Y = data['Y'][:, 0:1]
|
||||
Y[Y.flatten()==-1] = 0
|
||||
|
||||
# Kernel object
|
||||
kernel = GPy.kern.rbf(12)
|
||||
|
||||
# Likelihood object
|
||||
distribution = GPy.likelihoods.likelihood_functions.probit()
|
||||
likelihood = GPy.likelihoods.EP(data['Y'][:, 0:1], distribution)
|
||||
distribution = GPy.likelihoods.likelihood_functions.binomial()
|
||||
likelihood = GPy.likelihoods.EP(Y, distribution)
|
||||
|
||||
# Create GP model
|
||||
m = GPy.models.GP(data['X'], likelihood=likelihood, kernel=kernel)
|
||||
|
|
@ -79,12 +84,14 @@ def toy_linear_1d_classification(seed=default_seed):
|
|||
|
||||
data = GPy.util.datasets.toy_linear_1d_classification(seed=seed)
|
||||
Y = data['Y'][:, 0:1]
|
||||
Y[Y.flatten() == -1] = 0
|
||||
|
||||
# Kernel object
|
||||
kernel = GPy.kern.rbf(1)
|
||||
|
||||
# Likelihood object
|
||||
distribution = GPy.likelihoods.likelihood_functions.probit()
|
||||
link = GPy.likelihoods.link_functions.probit
|
||||
distribution = GPy.likelihoods.likelihood_functions.binomial(link)
|
||||
likelihood = GPy.likelihoods.EP(Y, distribution)
|
||||
|
||||
# Model definition
|
||||
|
|
@ -115,12 +122,13 @@ def sparse_toy_linear_1d_classification(seed=default_seed):
|
|||
|
||||
data = GPy.util.datasets.toy_linear_1d_classification(seed=seed)
|
||||
Y = data['Y'][:, 0:1]
|
||||
Y[Y.flatten() == -1] = 0
|
||||
|
||||
# Kernel object
|
||||
kernel = GPy.kern.rbf(1) + GPy.kern.white(1)
|
||||
|
||||
# Likelihood object
|
||||
distribution = GPy.likelihoods.likelihood_functions.probit()
|
||||
distribution = GPy.likelihoods.likelihood_functions.binomial()
|
||||
likelihood = GPy.likelihoods.EP(Y, distribution)
|
||||
|
||||
Z = np.random.uniform(data['X'].min(), data['X'].max(), (10, 1))
|
||||
|
|
@ -156,13 +164,15 @@ def sparse_crescent_data(inducing=10, seed=default_seed):
|
|||
"""
|
||||
|
||||
data = GPy.util.datasets.crescent_data(seed=seed)
|
||||
Y = data['Y']
|
||||
Y[Y.flatten()==-1]=0
|
||||
|
||||
# Kernel object
|
||||
kernel = GPy.kern.rbf(data['X'].shape[1]) + GPy.kern.white(data['X'].shape[1])
|
||||
|
||||
# Likelihood object
|
||||
distribution = GPy.likelihoods.likelihood_functions.probit()
|
||||
likelihood = GPy.likelihoods.EP(data['Y'], distribution)
|
||||
distribution = GPy.likelihoods.likelihood_functions.binomial()
|
||||
likelihood = GPy.likelihoods.EP(Y, distribution)
|
||||
|
||||
sample = np.random.randint(0, data['X'].shape[0], inducing)
|
||||
Z = data['X'][sample, :]
|
||||
|
|
|
|||
|
|
@ -7,6 +7,7 @@ from matplotlib import pyplot as plt
|
|||
import GPy
|
||||
from GPy.models.Bayesian_GPLVM import Bayesian_GPLVM
|
||||
from GPy.util.datasets import swiss_roll_generated
|
||||
from GPy.core.transformations import logexp
|
||||
|
||||
default_seed = np.random.seed(123344)
|
||||
|
||||
|
|
@ -17,11 +18,11 @@ def BGPLVM(seed=default_seed):
|
|||
D = 4
|
||||
# generate GPLVM-like data
|
||||
X = np.random.rand(N, Q)
|
||||
k = GPy.kern.rbf(Q) + GPy.kern.white(Q, 0.00001)
|
||||
k = GPy.kern.rbf(Q) + GPy.kern.white(Q, 0.00001)
|
||||
K = k.K(X)
|
||||
Y = np.random.multivariate_normal(np.zeros(N), K, D).T
|
||||
|
||||
k = GPy.kern.rbf(Q, ARD=True) + GPy.kern.linear(Q, ARD=True) + GPy.kern.rbf(Q, ARD=True) + GPy.kern.white(Q)
|
||||
k = GPy.kern.rbf(Q, ARD=True) + GPy.kern.linear(Q, ARD=True) + GPy.kern.rbf(Q, ARD=True) + GPy.kern.white(Q)
|
||||
# k = GPy.kern.rbf(Q) + GPy.kern.rbf(Q) + GPy.kern.white(Q)
|
||||
# k = GPy.kern.rbf(Q) + GPy.kern.bias(Q) + GPy.kern.white(Q, 0.00001)
|
||||
# k = GPy.kern.rbf(Q, ARD = False) + GPy.kern.white(Q, 0.00001)
|
||||
|
|
@ -187,8 +188,8 @@ def _simulate_sincos(D1, D2, D3, N, M, Q, plot_sim=False):
|
|||
Y2 = S2.dot(np.random.randn(S2.shape[1], D2))
|
||||
Y3 = S3.dot(np.random.randn(S3.shape[1], D3))
|
||||
|
||||
Y1 += .1 * np.random.randn(*Y1.shape)
|
||||
Y2 += .1 * np.random.randn(*Y2.shape)
|
||||
Y1 += .3 * np.random.randn(*Y1.shape)
|
||||
Y2 += .2 * np.random.randn(*Y2.shape)
|
||||
Y3 += .1 * np.random.randn(*Y3.shape)
|
||||
|
||||
Y1 -= Y1.mean(0)
|
||||
|
|
@ -262,13 +263,13 @@ def bgplvm_simulation(optimize='scg',
|
|||
# m.constrain('variance|noise', logexp_clipped())
|
||||
m.ensure_default_constraints()
|
||||
m['noise'] = Y.var() / 100.
|
||||
m['linear_variance'] = .01
|
||||
m['linear_variance'] = .001
|
||||
|
||||
if optimize:
|
||||
print "Optimizing model:"
|
||||
m.optimize('bfgs', max_iters=max_f_eval,
|
||||
m.optimize('scg', max_iters=max_f_eval,
|
||||
max_f_eval=max_f_eval,
|
||||
messages=True, gtol=1e-2)
|
||||
messages=True, gtol=1e-6)
|
||||
if plot:
|
||||
import pylab
|
||||
m.plot_X_1d()
|
||||
|
|
@ -277,23 +278,21 @@ def bgplvm_simulation(optimize='scg',
|
|||
m.kern.plot_ARD()
|
||||
return m
|
||||
|
||||
def mrd_simulation(optimize=True, plot_sim=False, **kw):
|
||||
D1, D2, D3, N, M, Q = 150, 200, 400, 300, 3, 7
|
||||
def mrd_simulation(optimize=True, plot=True, plot_sim=True, **kw):
|
||||
D1, D2, D3, N, M, Q = 150, 200, 400, 500, 3, 7
|
||||
slist, Slist, Ylist = _simulate_sincos(D1, D2, D3, N, M, Q, plot_sim)
|
||||
|
||||
from GPy.models import mrd
|
||||
from GPy import kern
|
||||
from GPy.core.transformations import logexp_clipped
|
||||
|
||||
reload(mrd); reload(kern)
|
||||
|
||||
k = kern.linear(Q, [0.05] * Q, True) + kern.bias(Q, np.exp(-2)) + kern.white(Q, np.exp(-2))
|
||||
m = mrd.MRD(Ylist, Q=Q, M=M, kernels=k, initx="concat", initz='permute', **kw)
|
||||
k = kern.linear(Q, [.05] * Q, ARD=True) + kern.bias(Q, np.exp(-2)) + kern.white(Q, np.exp(-2))
|
||||
m = mrd.MRD(Ylist, Q=Q, M=M, kernels=k, initx="", initz='permute', **kw)
|
||||
|
||||
for i, Y in enumerate(Ylist):
|
||||
m['{}_noise'.format(i + 1)] = Y.var() / 100.
|
||||
|
||||
# m.constrain('variance|noise', logexp_clipped(1e-6))
|
||||
m.ensure_default_constraints()
|
||||
|
||||
# DEBUG
|
||||
|
|
@ -301,8 +300,10 @@ def mrd_simulation(optimize=True, plot_sim=False, **kw):
|
|||
|
||||
if optimize:
|
||||
print "Optimizing Model:"
|
||||
m.optimize('bfgs', messages=1, max_iters=3e3)
|
||||
|
||||
m.optimize('scg', messages=1, max_iters=5e4, max_f_eval=5e4)
|
||||
if plot:
|
||||
m.plot_X_1d()
|
||||
m.plot_scales()
|
||||
return m
|
||||
|
||||
def brendan_faces():
|
||||
|
|
@ -323,7 +324,7 @@ def brendan_faces():
|
|||
m.ensure_default_constraints()
|
||||
m.optimize('scg', messages=1, max_f_eval=10000)
|
||||
|
||||
ax = m.plot_latent(which_indices=(0,1))
|
||||
ax = m.plot_latent(which_indices=(0, 1))
|
||||
y = m.likelihood.Y[0, :]
|
||||
data_show = GPy.util.visualize.image_show(y[None, :], dimensions=(20, 28), transpose=True, invert=False, scale=False)
|
||||
lvm_visualizer = GPy.util.visualize.lvm(m.X[0, :].copy(), m, data_show, ax)
|
||||
|
|
|
|||
|
|
@ -10,7 +10,7 @@ import numpy as np
|
|||
import GPy
|
||||
|
||||
|
||||
def toy_rbf_1d():
|
||||
def toy_rbf_1d(max_nb_eval_optim=100):
|
||||
"""Run a simple demonstration of a standard Gaussian process fitting it to data sampled from an RBF covariance."""
|
||||
data = GPy.util.datasets.toy_rbf_1d()
|
||||
|
||||
|
|
@ -19,13 +19,13 @@ def toy_rbf_1d():
|
|||
|
||||
# optimize
|
||||
m.ensure_default_constraints()
|
||||
m.optimize()
|
||||
m.optimize(max_f_eval=max_nb_eval_optim)
|
||||
# plot
|
||||
m.plot()
|
||||
print(m)
|
||||
return m
|
||||
|
||||
def rogers_girolami_olympics():
|
||||
def rogers_girolami_olympics(max_nb_eval_optim=100):
|
||||
"""Run a standard Gaussian process regression on the Rogers and Girolami olympics data."""
|
||||
data = GPy.util.datasets.rogers_girolami_olympics()
|
||||
|
||||
|
|
@ -34,14 +34,14 @@ def rogers_girolami_olympics():
|
|||
|
||||
# optimize
|
||||
m.ensure_default_constraints()
|
||||
m.optimize()
|
||||
m.optimize(max_f_eval=max_nb_eval_optim)
|
||||
|
||||
# plot
|
||||
m.plot(plot_limits = (1850, 2050))
|
||||
print(m)
|
||||
return m
|
||||
|
||||
def toy_rbf_1d_50():
|
||||
def toy_rbf_1d_50(max_nb_eval_optim=100):
|
||||
"""Run a simple demonstration of a standard Gaussian process fitting it to data sampled from an RBF covariance."""
|
||||
data = GPy.util.datasets.toy_rbf_1d_50()
|
||||
|
||||
|
|
@ -50,14 +50,14 @@ def toy_rbf_1d_50():
|
|||
|
||||
# optimize
|
||||
m.ensure_default_constraints()
|
||||
m.optimize()
|
||||
m.optimize(max_f_eval=max_nb_eval_optim)
|
||||
|
||||
# plot
|
||||
m.plot()
|
||||
print(m)
|
||||
return m
|
||||
|
||||
def silhouette():
|
||||
def silhouette(max_nb_eval_optim=100):
|
||||
"""Predict the pose of a figure given a silhouette. This is a task from Agarwal and Triggs 2004 ICML paper."""
|
||||
data = GPy.util.datasets.silhouette()
|
||||
|
||||
|
|
@ -66,12 +66,12 @@ def silhouette():
|
|||
|
||||
# optimize
|
||||
m.ensure_default_constraints()
|
||||
m.optimize(messages=True)
|
||||
m.optimize(messages=True,max_f_eval=max_nb_eval_optim)
|
||||
|
||||
print(m)
|
||||
return m
|
||||
|
||||
def coregionalisation_toy2():
|
||||
def coregionalisation_toy2(max_nb_eval_optim=100):
|
||||
"""
|
||||
A simple demonstration of coregionalisation on two sinusoidal functions.
|
||||
"""
|
||||
|
|
@ -90,8 +90,7 @@ def coregionalisation_toy2():
|
|||
m.constrain_fixed('rbf_var',1.)
|
||||
m.constrain_positive('kappa')
|
||||
m.ensure_default_constraints()
|
||||
m.optimize('sim',max_f_eval=5000,messages=1)
|
||||
#m.optimize()
|
||||
m.optimize('sim',messages=1,max_f_eval=max_nb_eval_optim)
|
||||
|
||||
pb.figure()
|
||||
Xtest1 = np.hstack((np.linspace(0,9,100)[:,None],np.zeros((100,1))))
|
||||
|
|
@ -104,7 +103,7 @@ def coregionalisation_toy2():
|
|||
pb.plot(X2[:,0],Y2[:,0],'gx',mew=2)
|
||||
return m
|
||||
|
||||
def coregionalisation_toy():
|
||||
def coregionalisation_toy(max_nb_eval_optim=100):
|
||||
"""
|
||||
A simple demonstration of coregionalisation on two sinusoidal functions.
|
||||
"""
|
||||
|
|
@ -123,7 +122,7 @@ def coregionalisation_toy():
|
|||
m.constrain_fixed('rbf_var',1.)
|
||||
m.constrain_positive('kappa')
|
||||
m.ensure_default_constraints()
|
||||
m.optimize()
|
||||
m.optimize(max_f_eval=max_nb_eval_optim)
|
||||
|
||||
pb.figure()
|
||||
Xtest1 = np.hstack((np.linspace(0,9,100)[:,None],np.zeros((100,1))))
|
||||
|
|
@ -137,7 +136,7 @@ def coregionalisation_toy():
|
|||
return m
|
||||
|
||||
|
||||
def coregionalisation_sparse():
|
||||
def coregionalisation_sparse(max_nb_eval_optim=100):
|
||||
"""
|
||||
A simple demonstration of coregionalisation on two sinusoidal functions using sparse approximations.
|
||||
"""
|
||||
|
|
@ -162,7 +161,7 @@ def coregionalisation_sparse():
|
|||
m.constrain_positive('kappa')
|
||||
m.constrain_fixed('iip')
|
||||
m.ensure_default_constraints()
|
||||
m.optimize_restarts(5,robust=True,messages=1)
|
||||
m.optimize_restarts(5, robust=True, messages=1, max_f_eval=max_nb_eval_optim)
|
||||
|
||||
pb.figure()
|
||||
Xtest1 = np.hstack((np.linspace(0,9,100)[:,None],np.zeros((100,1))))
|
||||
|
|
@ -179,7 +178,7 @@ def coregionalisation_sparse():
|
|||
return m
|
||||
|
||||
|
||||
def multiple_optima(gene_number=937,resolution=80, model_restarts=10, seed=10000):
|
||||
def multiple_optima(gene_number=937,resolution=80, model_restarts=10, seed=10000, max_nb_eval_optim=100):
|
||||
"""Show an example of a multimodal error surface for Gaussian process regression. Gene 939 has bimodal behaviour where the noisey mode is higher."""
|
||||
|
||||
# Contour over a range of length scales and signal/noise ratios.
|
||||
|
|
@ -217,7 +216,7 @@ def multiple_optima(gene_number=937,resolution=80, model_restarts=10, seed=10000
|
|||
|
||||
# optimize
|
||||
m.ensure_default_constraints()
|
||||
m.optimize(xtol=1e-6,ftol=1e-6)
|
||||
m.optimize(xtol=1e-6, ftol=1e-6, max_f_eval=max_nb_eval_optim)
|
||||
|
||||
optim_point_x[1] = m.get('rbf_lengthscale')
|
||||
optim_point_y[1] = np.log10(m.get('rbf_variance')) - np.log10(m.get('white_variance'));
|
||||
|
|
@ -264,7 +263,7 @@ def _contour_data(data, length_scales, log_SNRs, signal_kernel_call=GPy.kern.rbf
|
|||
lls.append(length_scale_lls)
|
||||
return np.array(lls)
|
||||
|
||||
def sparse_GP_regression_1D(N = 400, M = 5):
|
||||
def sparse_GP_regression_1D(N = 400, M = 5, max_nb_eval_optim=100):
|
||||
"""Run a 1D example of a sparse GP regression."""
|
||||
# sample inputs and outputs
|
||||
X = np.random.uniform(-3.,3.,(N,1))
|
||||
|
|
@ -279,11 +278,11 @@ def sparse_GP_regression_1D(N = 400, M = 5):
|
|||
m.constrain_positive('(variance|lengthscale|precision)')
|
||||
|
||||
m.checkgrad(verbose=1)
|
||||
m.optimize('tnc', messages = 1)
|
||||
m.optimize('tnc', messages = 1, max_f_eval=max_nb_eval_optim)
|
||||
m.plot()
|
||||
return m
|
||||
|
||||
def sparse_GP_regression_2D(N = 400, M = 50):
|
||||
def sparse_GP_regression_2D(N = 400, M = 50, max_nb_eval_optim=100):
|
||||
"""Run a 2D example of a sparse GP regression."""
|
||||
X = np.random.uniform(-3.,3.,(N,2))
|
||||
Y = np.sin(X[:,0:1]) * np.sin(X[:,1:2])+np.random.randn(N,1)*0.05
|
||||
|
|
@ -294,7 +293,7 @@ def sparse_GP_regression_2D(N = 400, M = 50):
|
|||
kernel = rbf + noise
|
||||
|
||||
# create simple GP model
|
||||
m = GPy.models.sparse_GP_regression(X,Y,kernel, M = M)
|
||||
m = GPy.models.sparse_GP_regression(X,Y,kernel, M = M, max_nb_eval_optim=100)
|
||||
|
||||
# contrain all parameters to be positive (but not inducing inputs)
|
||||
m.constrain_positive('(variance|lengthscale|precision)')
|
||||
|
|
@ -304,12 +303,12 @@ def sparse_GP_regression_2D(N = 400, M = 50):
|
|||
|
||||
# optimize and plot
|
||||
pb.figure()
|
||||
m.optimize('tnc', messages = 1)
|
||||
m.optimize('tnc', messages = 1, max_f_eval=max_nb_eval_optim)
|
||||
m.plot()
|
||||
print(m)
|
||||
return m
|
||||
|
||||
def uncertain_inputs_sparse_regression():
|
||||
def uncertain_inputs_sparse_regression(max_nb_eval_optim=100):
|
||||
"""Run a 1D example of a sparse GP regression with uncertain inputs."""
|
||||
# sample inputs and outputs
|
||||
S = np.ones((20,1))
|
||||
|
|
@ -327,7 +326,7 @@ def uncertain_inputs_sparse_regression():
|
|||
m.constrain_positive('(variance|prec)')
|
||||
|
||||
# optimize and plot
|
||||
m.optimize('tnc', max_f_eval = 1000, messages=1)
|
||||
m.optimize('tnc', messages=1, max_f_eval=max_nb_eval_optim)
|
||||
m.plot()
|
||||
print(m)
|
||||
return m
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue