Repository navigation
FilesExpand file tree
/
Copy path_sofia_ml.pyx
More file actions
94 lines (75 loc) · 3.59 KB
/
Copy path_sofia_ml.pyx
File metadata and controls
94 lines (75 loc) · 3.59 KB
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
# distutils: language = c++
# distutils: sources = minirank/src/sofia-ml-methods.cc ranking/src/{sf-weight-vector.cc,sf-data-set.cc}
from cython.operator cimport dereference as deref
from libcpp cimport bool
from libcpp.string cimport string
from libcpp.vector cimport vector
cimport numpy as np
import numpy as np
BUFFER_MB = 40 # default in sofia-ml
cdef extern from "src/sofia-ml-methods.h":
cdef cppclass SfDataSet:
SfDataSet(bool)
SfDataSet(string, int, bool)
cdef cppclass SfWeightVector:
SfWeightVector(int)
SfWeightVector(string)
string AsString()
float ValueOf(int)
cdef extern from "src/sofia-ml-methods.h" namespace "sofia_ml":
cdef enum LearnerType:
PEGASOS, MARGIN_PERCEPTRON, PASSIVE_AGGRESSIVE, LOGREG_PEGASOS,
LOGREG, LMS_REGRESSION, SGD_SVM, ROMMA
cdef enum EtaType:
BASIC_ETA
PEGASOS_ETA
CONSTANT
void StochasticRocLoop(SfDataSet, LearnerType, EtaType,
float, float, int, SfWeightVector*)
void BalancedStochasticOuterLoop(SfDataSet, LearnerType, EtaType,
float, float, int, SfWeightVector*)
void StochasticRankLoop(SfDataSet, LearnerType, EtaType,
float, float, int, SfWeightVector*)
void StochasticClassificationAndRankLoop(SfDataSet, LearnerType, EtaType,
float, float, float, int num_iters, SfWeightVector*)
void SvmPredictionsOnTestSet(SfDataSet test_data,
SfWeightVector, vector[float]*)
void StochasticOuterLoop(SfDataSet, LearnerType, EtaType,
float, float, int num_iters, SfWeightVector*)
void StochasticQueryNormRankLoop(SfDataSet, LearnerType, EtaType,
float, float, int num_iters, SfWeightVector*)
def train(train_data, int n_features, float alpha, int max_iter, bool fit_intercept,
model, float step_probability):
cdef SfDataSet *data = new SfDataSet(train_data, BUFFER_MB, fit_intercept)
cdef SfWeightVector *w = new SfWeightVector(n_features)
cdef float c = 0.0
cdef int i
# TODO : Select learner_type and eta_type. Rename model to loop_type.
if model == 'rank':
StochasticRankLoop(deref(data), PEGASOS, BASIC_ETA, alpha, c, max_iter, w)
elif model == 'roc':
StochasticRocLoop(deref(data), SGD_SVM, BASIC_ETA, alpha, c, max_iter, w)
elif model == 'combined-ranking':
StochasticClassificationAndRankLoop(deref(data), SGD_SVM, BASIC_ETA, alpha, c,
step_probability, max_iter, w)
elif model == 'stochastic':
StochasticOuterLoop(deref(data), SGD_SVM, BASIC_ETA, alpha, c, max_iter, w)
elif model == 'balanced-stochastic':
BalancedStochasticOuterLoop(deref(data), SGD_SVM, BASIC_ETA, alpha, c, max_iter, w)
elif model == 'query-norm-rank':
StochasticQueryNormRankLoop(deref(data), SGD_SVM, BASIC_ETA, alpha, c, max_iter, w)
else:
raise NotImplementedError
cdef np.ndarray[ndim=1, dtype=np.float64_t] coef = np.empty(n_features)
for i in range(n_features):
coef[i] = w.ValueOf(i)
return coef
def predict(test_data, string coef, bool fit_intercept):
cdef SfDataSet *test_dataset = new SfDataSet(test_data, BUFFER_MB, fit_intercept)
cdef SfWeightVector *w = new SfWeightVector(coef)
cdef vector[float] *predictions = new vector[float]()
SvmPredictionsOnTestSet(deref(test_dataset), deref(w), predictions)
cdef np.ndarray[ndim=1, dtype=np.float64_t] out = np.empty(predictions.size())
for i in range(predictions.size()):
out[i] = predictions.at(i)
return out