apunkt/GPy
1
0
Fork 0
mirror of https://github.com/SheffieldML/GPy.git synced 2026-05-15 06:52:39 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Polynomial kernel gradients wrt X working.

Browse source
This commit is contained in:
Neil Lawrence 2013-08-28 00:20:12 +02:00
parent a61feb17a5
commit b127c96bf2
8 changed files with 79 additions and 26 deletions
Download patch file Download diff file Expand all files Collapse all files

21
GPy/kern/parts/mlp.py
View file

@ -107,12 +107,22 @@ class MLP(Kernpart):
def dK_dX(self, dL_dK, X, X2, target):
"""Derivative of the covariance matrix with respect to X"""
self._K_computations(X, X2)
arg = self._K_asin_arg
post_div = np.sqrt(1-arg*arg)
numer = self._K_numer
denom = self._K_denom
vec2 = (X2*X2).sum(1)*self.weight_variance + self.bias_variance + 1.
denom3 = denom*denom*denom
target += (((X2[None,:, :]/denom[None, :, None]+vec2[None, None, :]*X[:, :, None]*numer/denom)/post_div[:, :, None]) * dL_dK[:, :, None]).sum(1)
target *= four_over_tau*self.weight_variance*self.variance
raise NotImplementedError
# self._K_computations(X, X2)
# gX = np.zeros((X2.shape[0], X.shape[1], X.shape[0]))
gX = np.zeros((X2.shape[0], X.shape[1], X.shape[0]))
# for i in range(X.shape[0]):
# gX[:, :, i] = self._dK_dX_point(dL_dK, X, X2, target, i)
for i in range(X.shape[0]):
gX[:, :, i] = self._dK_dX_point(dL_dK, X, X2, target, i)
def _dK_dX_point(self, dL_dK, X, X2, target, i):
@ -130,9 +140,8 @@ class MLP(Kernpart):
denom3 = denom*denom*denom
gX = np.zeros((X2.shape[0], X2.shape[1]))
for j in range(X2.shape[1]):
gX[:, j] =X2[:, j]/denom - vec2*X[i, j]*numer/denom3
gX[:, j] = X2[:, j]/denom - vec2*X[i, j]*numer/denom3
gX[:, j] = four_over_tau*self.weight_variance*self.variance*gX[:, j]/np.sqrt(1-arg*arg)
target[i, :]
def _K_computations(self, X, X2):

12
GPy/kern/parts/poly.py
View file

@ -101,11 +101,15 @@ class POLY(Kernpart):
def dK_dX(self, dL_dK, X, X2, target):
"""Derivative of the covariance matrix with respect to X"""
pass
self._K_computations(X, X2)
arg = self._K_poly_arg
target += self.weight_variance*self.degree*self.variance*(((X2[None,:, :])) *(arg**(self.degree-1))[:, :, None]* dL_dK[:, :, None]).sum(1)
def _dK_dX_point(self, dL_dK, X, X2, target, i):
"""Gradient with respect to one point of X"""
pass
def dKdiag_dX(self, dL_dKdiag, X, target):
"""Gradient of diagonal of covariance with respect to X"""
self._K_diag_computations(X)
arg = self._K_diag_poly_arg
target += 2.*self.weight_variance*self.degree*self.variance*X*dL_dKdiag[:, None]*(arg**(self.degree-1))[:, None]
def _K_computations(self, X, X2):

2
GPy/kern/parts/prod.py
View file

@ -62,7 +62,7 @@ class Prod(Kernpart):
return self._K2
def dK_dtheta(self,dL_dK,X,X2,target):
"""derivative of the covariance matrix with respect to the parameters."""
"""Derivative of the covariance matrix with respect to the parameters."""
self._K_computations(X,X2)
if X2 is None:
self.k1.dK_dtheta(dL_dK*self._K2, X[:,self.slice1], None, target[:self.k1.num_params])