-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathlibopf_py.pyx
241 lines (195 loc) · 8.99 KB
/
libopf_py.pyx
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
# code
cimport libopf_py
import numpy as np
cimport numpy as np
d = {
"euclidian" : libopf_py.EUCLIDIAN,
"log_euclidian" : libopf_py.LOG_EUCLIDIAN,
"chi_square" : libopf_py.CHI_SQUARE,
"manhattan" : libopf_py.MANHATTAN,
"canberra" : libopf_py.CANBERRA,
"squared_chord" : libopf_py.SQUARED_CHORD,
"squared_chi_square" : libopf_py.SQUARED_CHI_SQUARE,
"bray_curtis" : libopf_py.BRAY_CURTIS
}
cdef class OPF:
cdef libopf_py.opf_graph * sg
cdef bint supervised
cdef bint precomputed_distance
cdef int node_n
cdef int feat_n
cdef str metric
def __cinit__(self):
self.sg = NULL
self.node_n = 0
self.feat_n = 0
self.supervised = True
self.precomputed_distance = False
self.metric = "euclidian"
def __dealloc__(self):
if self.sg is not NULL:
libopf_py.opf_graph_destroy (&self.sg)
def __reduce__(self):
cdef np.ndarray[np.float64_t, ndim=1, mode='c'] path_val, radius
cdef np.ndarray[np.float64_t, ndim=2, mode='c'] X
cdef np.ndarray[np.int32_t, ndim=1, mode='c'] ordered_list_of_nodes, position, label
path_val = np.empty(self.node_n)
label = np.empty(self.node_n, dtype=np.int32)
ordered_list_of_nodes = np.empty(self.node_n, dtype=np.int32)
position = np.empty(self.node_n, dtype=np.int32)
radius = np.empty(self.node_n)
if not self.precomputed_distance:
X = np.empty((self.node_n, self.feat_n))
else:
X = np.empty(0)
libopf_py.opf_graph_get_fit_data (self.sg,
<double *>path_val.data,
<int *>label.data,
<int *>ordered_list_of_nodes.data,
<int *>position.data,
<double *>radius.data,
<double *>X.data)
return OPF_unpickle, (self.node_n, self.feat_n, self.metric, self.supervised,
self.precomputed_distance, path_val, label, ordered_list_of_nodes,
position, radius, X)
def fit (self,
np.ndarray[np.float64_t, ndim=2, mode='c'] X,
np.ndarray[np.int32_t, ndim=1, mode='c'] Y = None,
learning="default", metric="euclidian",
bint precomputed_distance=False, double split=0.8):
if Y != None:
self.supervised = True
else:
self.supervised = False
self.precomputed_distance = precomputed_distance
self.node_n = <int>X.shape[0]
self.feat_n = <int>X.shape[1]
self.metric = metric
if Y != None and X.shape[0] != Y.shape[0]:
raise Exception("Shape mismatch")
if self.precomputed_distance and X.shape[0] != X.shape[1]:
raise Exception("Distance matrix should be squared, but it's (%s,%s)" %
(X.shape[0], X.shape[1]))
if self.supervised:
if learning not in ("default", "iterative", "agglomerative"):
raise Exception("Invalid training mode")
self.sg = libopf_py.opf_graph_create (<int>X.shape[0])
if self.sg == NULL:
raise MemoryError("Seems we've run out of of memory")
if self.precomputed_distance:
if self.supervised:
if not libopf_py.opf_graph_set_precomputed_distance (self.sg,
<double*>X.data,
<int*>Y.data):
raise MemoryError("Seems we've run out of of memory")
else:
if not libopf_py.opf_graph_set_precomputed_distance (self.sg,
<double*>X.data,
NULL):
raise MemoryError("Seems we've run out of of memory")
else:
if self.supervised:
if not libopf_py.opf_graph_set_feature (self.sg,
<double*>X.data,
<int*>Y.data,
<int>X.shape[1]):
raise MemoryError("Seems we've run out of of memory")
else:
if not libopf_py.opf_graph_set_feature (self.sg,
<double*>X.data,
NULL,
<int>X.shape[1]):
raise MemoryError("Seems we've run out of of memory")
libopf_py.opf_graph_set_metric (self.sg, NULL, d[metric])
if self.supervised:
if learning == "default":
libopf_py.opf_supervised_train (self.sg)
elif learning == "iterative":
libopf_py.opf_supervised_train_iterative (self.sg, split)
elif learning == "agglomerative":
libopf_py.opf_supervised_train_agglomerative (self.sg, split)
else:
libopf_py.opf_best_k_min_cut (self.sg, 1, 10)
libopf_py.opf_unsupervised_clustering (self.sg)
def predict(self, np.ndarray[np.float64_t, ndim=2, mode='c'] X):
cdef np.ndarray[np.int32_t, ndim=1, mode='c'] labels
if self.precomputed_distance:
labels = np.empty(X.shape[1], dtype=np.int32)
else:
labels = np.empty(X.shape[0], dtype=np.int32)
if self.supervised == None:
raise Exception ("Not fitted!")
if self.precomputed_distance and X.shape[0] != self.node_n:
raise Exception("Distance matrix shape is wrong")
if not self.precomputed_distance and X.shape[1] != self.feat_n:
raise Exception("Feature matrix shape is wrong")
if self.precomputed_distance:
if self.supervised:
libopf_py.opf_supervised_classify (self.sg,
<double*>X.data,
<int>X.shape[1],
<int*>labels.data)
else:
libopf_py.opf_unsupervised_knn_classify (self.sg,
<double*>X.data,
<int>X.shape[1],
<int*>labels.data)
else:
if self.supervised:
libopf_py.opf_supervised_classify (self.sg,
<double*>X.data,
<int>X.shape[0],
<int*>labels.data)
else:
libopf_py.opf_unsupervised_knn_classify (self.sg,
<double*>X.data,
<int>X.shape[0],
<int*>labels.data)
return labels
def OPF_unpickle (*args):
cdef np.ndarray[np.float64_t, ndim=1, mode='c'] path_val, radius
cdef np.ndarray[np.float64_t, ndim=2, mode='c'] X
cdef np.ndarray[np.int32_t, ndim=1, mode='c'] Y, ordered_list_of_nodes, position
cdef OPF opf
node_n, feat_n, metric, supervised, precomputed_distance, \
path_val, Y, ordered_list_of_nodes, position, \
radius, X = args
opf = OPF()
opf.node_n = node_n
opf.feat_n = feat_n
opf.metric = metric
opf.supervised = supervised
opf.precomputed_distance = precomputed_distance
opf.sg = libopf_py.opf_graph_create (opf.node_n)
if opf.precomputed_distance:
if opf.supervised:
if not libopf_py.opf_graph_set_precomputed_distance (opf.sg,
<double*>X.data,
<int*>Y.data):
raise MemoryError("Seems we've run out of of memory")
else:
if not libopf_py.opf_graph_set_precomputed_distance (opf.sg,
<double*>X.data,
NULL):
raise MemoryError("Seems we've run out of of memory")
else:
if opf.supervised:
if not libopf_py.opf_graph_set_feature (opf.sg,
<double*>X.data,
<int*>Y.data,
<int>X.shape[1]):
raise MemoryError("Seems we've run out of of memory")
else:
if not libopf_py.opf_graph_set_feature (opf.sg,
<double*>X.data,
NULL,
<int>X.shape[1]):
raise MemoryError("Seems we've run out of of memory")
libopf_py.opf_graph_set_metric (opf.sg, NULL, d[metric])
libopf_py.opf_graph_set_fit_data (opf.sg,
<double *>path_val.data,
<int *>Y.data,
<int *>ordered_list_of_nodes.data,
<int *>position.data,
<double *>radius.data)
return opf