OpenI
/
graphkit-learn
Not watched
1
0
Fork 0
Code
Releases 2 Wiki Activity
Issues 0 Pull Requests 0 Datasets Model Cloudbrain
Browse Source

Fix bugs in four random walk kernels.

v0.2.x
jajupmochi 4 years ago
parent
00aee56118
commit
f05954f302
3 changed files with 28 additions and 28 deletions
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. +10
    -10
      gklearn/kernels/conjugate_gradient.py
  2. +11
    -11
      gklearn/kernels/fixed_point.py
  3. +7
    -7
      gklearn/kernels/spectral_decomposition.py

+ 10
- 10
gklearn/kernels/conjugate_gradient.py View File

@@ -60,7 +60,7 @@ class ConjugateGradient(RandomWalkMeta):
iterator = itr
for i, j in iterator:
kernel = self.__kernel_do(self._graphs[i], self._graphs[j], lmda)
kernel = self._kernel_do(self._graphs[i], self._graphs[j], lmda)
gram_matrix[i][j] = kernel
gram_matrix[j][i] = kernel

@@ -127,7 +127,7 @@ class ConjugateGradient(RandomWalkMeta):
iterator = range(len(g_list))
for i in iterator:
kernel = self.__kernel_do(g1, g_list[i], lmda)
kernel = self._kernel_do(g1, g_list[i], lmda)
kernel_list[i] = kernel

else: # @todo
@@ -190,7 +190,7 @@ class ConjugateGradient(RandomWalkMeta):
g2 = nx.convert_node_labels_to_integers(g2, first_label=0, label_attribute='label_orignal')
if self._p is None and self._q is None: # p and q are uniform distributions as default.
kernel = self.__kernel_do(g1, g2, lmda)
kernel = self._kernel_do(g1, g2, lmda)

else: # @todo
pass
@@ -198,7 +198,7 @@ class ConjugateGradient(RandomWalkMeta):
return kernel
def __kernel_do(self, g1, g2, lmda):
def _kernel_do(self, g1, g2, lmda):
# Frist, compute kernels between all pairs of nodes using the method borrowed
# from FCSP. It is faster than directly computing all edge kernels
@@ -222,7 +222,7 @@ class ConjugateGradient(RandomWalkMeta):
def _wrapper_kernel_do(self, itr):
i = itr[0]
j = itr[1]
return i, j, self.__kernel_do(G_gn[i], G_gn[j], self._weight)
return i, j, self._kernel_do(G_gn[i], G_gn[j], self._weight)
def _func_fp(x, p_times, lmda, w_times):
@@ -246,19 +246,19 @@ class ConjugateGradient(RandomWalkMeta):
# Define edge kernels.
def compute_ek_11(e1, e2, ke):
e1_labels = [e1[2][el] for el in self._edge_labels]
e2_labels = [e2[2][el] for el in self.__edge_labels]
e2_labels = [e2[2][el] for el in self._edge_labels]
e1_attrs = [e1[2][ea] for ea in self._edge_attrs]
e2_attrs = [e2[2][ea] for ea in self._edge_attrs]
return ke(e1_labels, e2_labels, e1_attrs, e2_attrs)
def compute_ek_10(e1, e2, ke):
e1_labels = [e1[2][el] for el in self.__edge_labels]
e2_labels = [e2[2][el] for el in self.__edge_labels]
e1_labels = [e1[2][el] for el in self._edge_labels]
e2_labels = [e2[2][el] for el in self._edge_labels]
return ke(e1_labels, e2_labels)
def compute_ek_01(e1, e2, ke):
e1_attrs = [e1[2][ea] for ea in self.__edge_attrs]
e2_attrs = [e2[2][ea] for ea in self.__edge_attrs]
e1_attrs = [e1[2][ea] for ea in self._edge_attrs]
e2_attrs = [e2[2][ea] for ea in self._edge_attrs]
return ke(e1_attrs, e2_attrs)
def compute_ek_00(e1, e2, ke):


+ 11
- 11
gklearn/kernels/fixed_point.py View File

@@ -60,7 +60,7 @@ class FixedPoint(RandomWalkMeta):
iterator = itr
for i, j in iterator:
kernel = self.__kernel_do(self._graphs[i], self._graphs[j], lmda)
kernel = self._kernel_do(self._graphs[i], self._graphs[j], lmda)
gram_matrix[i][j] = kernel
gram_matrix[j][i] = kernel

@@ -127,7 +127,7 @@ class FixedPoint(RandomWalkMeta):
iterator = range(len(g_list))
for i in iterator:
kernel = self.__kernel_do(g1, g_list[i], lmda)
kernel = self._kernel_do(g1, g_list[i], lmda)
kernel_list[i] = kernel

else: # @todo
@@ -190,7 +190,7 @@ class FixedPoint(RandomWalkMeta):
g2 = nx.convert_node_labels_to_integers(g2, first_label=0, label_attribute='label_orignal')
if self._p is None and self._q is None: # p and q are uniform distributions as default.
kernel = self.__kernel_do(g1, g2, lmda)
kernel = self._kernel_do(g1, g2, lmda)

else: # @todo
pass
@@ -198,7 +198,7 @@ class FixedPoint(RandomWalkMeta):
return kernel
def __kernel_do(self, g1, g2, lmda):
def _kernel_do(self, g1, g2, lmda):
# Frist, compute kernels between all pairs of nodes using the method borrowed
# from FCSP. It is faster than directly computing all edge kernels
@@ -221,10 +221,10 @@ class FixedPoint(RandomWalkMeta):
def _wrapper_kernel_do(self, itr):
i = itr[0]
j = itr[1]
return i, j, self.__kernel_do(G_gn[i], G_gn[j], self._weight)
return i, j, self._kernel_do(G_gn[i], G_gn[j], self._weight)
def _func_fp(x, p_times, lmda, w_times):
def _func_fp(self, x, p_times, lmda, w_times):
haha = w_times * x
haha = lmda * haha
haha = p_times + haha
@@ -245,19 +245,19 @@ class FixedPoint(RandomWalkMeta):
# Define edge kernels.
def compute_ek_11(e1, e2, ke):
e1_labels = [e1[2][el] for el in self._edge_labels]
e2_labels = [e2[2][el] for el in self.__edge_labels]
e2_labels = [e2[2][el] for el in self._edge_labels]
e1_attrs = [e1[2][ea] for ea in self._edge_attrs]
e2_attrs = [e2[2][ea] for ea in self._edge_attrs]
return ke(e1_labels, e2_labels, e1_attrs, e2_attrs)
def compute_ek_10(e1, e2, ke):
e1_labels = [e1[2][el] for el in self.__edge_labels]
e2_labels = [e2[2][el] for el in self.__edge_labels]
e1_labels = [e1[2][el] for el in self._edge_labels]
e2_labels = [e2[2][el] for el in self._edge_labels]
return ke(e1_labels, e2_labels)
def compute_ek_01(e1, e2, ke):
e1_attrs = [e1[2][ea] for ea in self.__edge_attrs]
e2_attrs = [e2[2][ea] for ea in self.__edge_attrs]
e1_attrs = [e1[2][ea] for ea in self._edge_attrs]
e2_attrs = [e2[2][ea] for ea in self._edge_attrs]
return ke(e1_attrs, e2_attrs)
def compute_ek_00(e1, e2, ke):


+ 7
- 7
gklearn/kernels/spectral_decomposition.py View File

@@ -66,7 +66,7 @@ class SpectralDecomposition(RandomWalkMeta):
iterator = itr
for i, j in iterator:
kernel = self.__kernel_do(q_T_list[i], q_T_list[j], P_list[i], P_list[j], D_list[i], D_list[j], self._weight, self._sub_kernel)
kernel = self._kernel_do(q_T_list[i], q_T_list[j], P_list[i], P_list[j], D_list[i], D_list[j], self._weight, self._sub_kernel)
gram_matrix[i][j] = kernel
gram_matrix[j][i] = kernel
@@ -162,7 +162,7 @@ class SpectralDecomposition(RandomWalkMeta):
iterator = range(len(g_list))
for i in iterator:
kernel = self.__kernel_do(q_T1, q_T_list[i], P1, P_list[i], D1, D_list[i], self._weight, self._sub_kernel)
kernel = self._kernel_do(q_T1, q_T_list[i], P1, P_list[i], D1, D_list[i], self._weight, self._sub_kernel)
kernel_list[i] = kernel
else: # @todo
@@ -190,9 +190,9 @@ class SpectralDecomposition(RandomWalkMeta):
P_list = []
D_list = []
if self._verbose >= 2:
iterator = tqdm(range(len(g_list)), desc='spectral decompose', file=sys.stdout)
iterator = tqdm(g_list, desc='spectral decompose', file=sys.stdout)
else:
iterator = range(len(g_list))
iterator = g_list
for G in iterator:
# don't normalize adjacency matrices if q is a uniform vector. Note
# A actually is the transpose of the adjacency matrix.
@@ -252,7 +252,7 @@ class SpectralDecomposition(RandomWalkMeta):
if self._p is None: # p is uniform distribution as default.
q_T1 = 1 / nx.number_of_nodes(g1)
q_T2 = 1 / nx.number_of_nodes(g2)
kernel = self.__kernel_do(q_T1, q_T2, P1, P2, D1, D2, self._weight, self._sub_kernel)
kernel = self._kernel_do(q_T1, q_T2, P1, P2, D1, D2, self._weight, self._sub_kernel)
else: # @todo
pass
else: # @todo
@@ -261,7 +261,7 @@ class SpectralDecomposition(RandomWalkMeta):
return kernel
def __kernel_do(self, q_T1, q_T2, P1, P2, D1, D2, weight, sub_kernel):
def _kernel_do(self, q_T1, q_T2, P1, P2, D1, D2, weight, sub_kernel):
# use uniform distribution if there is no prior knowledge.
kl = kron(np.dot(q_T1, P1), np.dot(q_T2, P2)).todense()
# @todo: this is not needed when p = q (kr = kl.T) for undirected graphs.
@@ -280,4 +280,4 @@ class SpectralDecomposition(RandomWalkMeta):
def _wrapper_kernel_do(self, itr):
i = itr[0]
j = itr[1]
return i, j, self.__kernel_do(G_q_T_list[i], G_q_T_list[j], G_P_list[i], G_P_list[j], G_D_list[i], G_D_list[j], self._weight, self._sub_kernel)
return i, j, self._kernel_do(G_q_T_list[i], G_q_T_list[j], G_P_list[i], G_P_list[j], G_D_list[i], G_D_list[j], self._weight, self._sub_kernel)

Write Preview
Loading…
Cancel
Save