apunkt/GPy
1
0
Fork 0
mirror of https://github.com/SheffieldML/GPy.git synced 2026-04-28 22:36:24 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

[beckdaniel] merge in warped gp changes

Browse source
This commit is contained in:
mzwiessele 2015-10-13 08:52:26 +01:00
commit 87539b1c1b
4 changed files with 183 additions and 57 deletions
Download patch file Download diff file Expand all files Collapse all files

58
GPy/testing/model_tests.py
View file

@ -289,6 +289,64 @@ class MiscTests(unittest.TestCase):
m.optimize()
print(m)
def test_warped_gp_identity(self):
"""
A WarpedGP with the identity warping function should be
equal to a standard GP.
"""
k = GPy.kern.RBF(1)
m = GPy.models.GPRegression(self.X, self.Y, kernel=k)
m.optimize()
preds = m.predict(self.X)
warp_k = GPy.kern.RBF(1)
warp_f = GPy.util.warping_functions.IdentityFunction()
warp_m = GPy.models.WarpedGP(self.X, self.Y, kernel=warp_k, warping_function=warp_f)
warp_m.optimize()
warp_preds = warp_m.predict(self.X)
np.testing.assert_almost_equal(preds, warp_preds)
@unittest.skip('Comment this to plot the modified sine function')
def test_warped_gp_sine(self):
"""
A test replicating the sine regression problem from
Snelson's paper.
"""
X = (2 * np.pi) * np.random.random(151) - np.pi
Y = np.sin(X) + np.random.normal(0,0.1,151)
Y = np.exp(Y) - 5
#Y = np.array([np.power(abs(y),float(1)/3) * (1,-1)[y<0] for y in Y]) + 0
#np.seterr(over='raise')
import matplotlib.pyplot as plt
warp_k = GPy.kern.RBF(1)
warp_f = GPy.util.warping_functions.TanhWarpingFunction_d(n_terms=2)
warp_m = GPy.models.WarpedGP(X[:, None], Y[:, None], kernel=warp_k, warping_function=warp_f)
#warp_m['.*variance.*'].constrain_fixed(0.25)
#warp_m['.*lengthscale.*'].constrain_fixed(1)
#warp_m['warp_tanh.d'].constrain_fixed(1)
#warp_m.randomize()
#warp_m['.*warp_tanh.psi*'][:,0:2].constrain_bounded(0,100)
#warp_m['.*warp_tanh.psi*'][:,0:1].constrain_fixed(1)
#print(warp_m.checkgrad())
#warp_m.plot()
#plt.show()
warp_m.optimize_restarts(parallel=True, robust=True)
#print(warp_m.checkgrad())
print(warp_m)
print(warp_m['.*warp.*'])
warp_m.predict_in_warped_space = False
warp_m.plot()
warp_m.predict_in_warped_space = True
warp_m.plot()
warp_f.plot(X.min()-10, X.max()+10)
plt.show()
class GradientTests(np.testing.TestCase):
def setUp(self):
######################################