-
Notifications
You must be signed in to change notification settings - Fork 0
/
embedded_space.py
305 lines (237 loc) · 11.5 KB
/
embedded_space.py
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
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
import mldata as mldata
import random
import copy
import numpy as np
import math
from sklearn.cluster import KMeans
from sklearn.ensemble import AdaBoostClassifier
import EDM54.miHelperMethods as utilities
import data_parser as dp
from EDM54.imk import IMK
from EDM54.dbow import DBOW
from EDM54.hbow import HBOW
from EDM54.yards import YARDS
from EDM54.simpleMi import SMI
import multiprocessing
import EDM54.similarities as similarities
import pandas as pd
std_list = ['imk', 'dbow', 'hbow', 'yards', 'smi']
class embeddedSpace:
def ES_MI(self, path, cross_validation, learning_algorithm, clusters=8, distance_metric='euclidean',
cluster_scheme='one', C=1):
if type(clusters) is list and len(clusters) == 1:
clusters = clusters[0]
# Standardize data
if learning_algorithm in std_list:
examples = dp.parse_data_file(path)
attribute_map = utilities.classify_attributes(path)
std_examples = utilities.standardize_examples(examples, attribute_map)
bags = dp.make_bags(std_examples)
else:
bags = dp.getBaggedData(path)
if cross_validation == 0:
self.cross_validation(bags, learning_algorithm, folds=10, iterations=2,
num_clusters=clusters, distance_metric = distance_metric, c=C)
elif learning_algorithm is 'dbow':
dbow = DBOW()
dbow.distance_bow(bags, clusters=clusters, distance_metric=distance_metric)
elif learning_algorithm is 'yards':
yards = YARDS()
yards.learn(bags)
elif learning_algorithm is 'noncluster':
self.nonclustering_BOW(bags)
elif learning_algorithm is 'imk':
imk = IMK()
imk.learn(bags, clusters=clusters)
elif learning_algorithm is 'hbow':
hbow = HBOW()
hbow.histogram_bow(bags, clusters=clusters, distance_metric=distance_metric)
else:
raise Exception('Error: Algorithm not recognized')
# use for fox or elephant
def ES_MI2(self, animal, cross_validation, learning_algorithm, clusters=8, distance_metric='euclidean',
cluster_scheme='one', C=1):
if type(clusters) is list and len(clusters) == 1:
clusters = clusters[0]
print(animal)
# Standardize data
if learning_algorithm in std_list:
if animal is 'elephant':
_, labels, data, l2, indexes = dp.loadelephant()
else:
_, labels, data, l2, indexes = dp.loadfox()
data = utilities.standardize_everything(data)
bags = dp.parse_svm(data, labels, indexes)
else:
if animal is 'elephant':
_, labels, data, l2, indexes = dp.loadelephant()
else:
_, labels, data, l2, indexes = dp.loadfox()
#bags = dp.getBaggedData(path)
bags = dp.parse_svm(data, labels, indexes)
if cross_validation == 0:
self.cross_validation(bags, learning_algorithm, folds=10,iterations=3,
num_clusters=clusters, distance_metric = distance_metric, c=C)
elif learning_algorithm is 'dbow':
dbow = DBOW()
dbow.distance_bow(bags, clusters=clusters, distance_metric=distance_metric)
elif learning_algorithm is 'yards':
yards = YARDS()
yards.learn(bags)
elif learning_algorithm is 'noncluster':
self.nonclustering_BOW(bags)
elif learning_algorithm is 'imk':
imk = IMK()
imk.learn(bags, clusters=clusters)
elif learning_algorithm is 'hbow':
hbow = HBOW()
hbow.histogram_bow(bags, clusters=clusters, distance_metric=distance_metric)
else:
raise Exception('Error: Algorithm not recognized')
def cluster_instances(self, instances, num_clusters = 8):
km = KMeans(n_clusters=num_clusters).fit(instances)
return km.cluster_centers_
def cross_validation(self, bags, learning_algorithm, folds=5, iterations=1, num_clusters=8,
distance_metric='euclidean',c=1):
bins = folds
class_one = []
class_two = []
accuracies = []
precisions = []
recalls = []
# find out how many in each class
for i in range(0, len(bags)):
if bags[i][-1]:
class_one.append(i)
elif not bags[i][-1]:
class_two.append(i)
# set the random seed
#np.random.seed(12345)
for iter in range(0,iterations):
# randomly shuffle the two
np.random.shuffle(class_one)
np.random.shuffle(class_two)
for a in range(0, bins):
testing_class_one = class_one[len(class_one) // bins * a:len(class_one) // bins * (a + 1)]
testing_class_two = class_two[len(class_two) // bins * a:len(class_two) // bins * (a + 1)]
training_class_one = [a for a in class_one if a not in testing_class_one]
training_class_two = [a for a in class_two if a not in testing_class_two]
all_class_labels_test = \
self.select_subset(bags, testing_class_one, testing_class_two)
all_class_labels_train = \
self.select_subset(bags, training_class_one, training_class_two)
#for a in all_class_labels_train:
#print(a[-1])
if learning_algorithm is 'dbow':
dbow = DBOW()
cluster_centers, classifier = dbow.distance_bow(all_class_labels_train, num_clusters,
distance_metric=distance_metric,c=c)
test_vector, test_labels = dbow.map_all(all_class_labels_test, cluster_centers,
num_clusters=num_clusters,
distance_metric=distance_metric)
results = classifier.predict(test_vector)
elif learning_algorithm is 'imk':
imk = IMK()
cluster_centers, classifier = imk.learn(all_class_labels_train, clusters=num_clusters, c=c)
test_vector, test_labels = imk.map_all_imk(all_class_labels_test, cluster_centers)
results = classifier.predict(test_vector)
elif learning_algorithm is 'hbow':
hbow = HBOW()
cluster_centers, classifier = hbow.histogram_bow(all_class_labels_train, num_clusters,
distance_metric=distance_metric, C=c)
test_vector, test_labels = hbow.map_all(all_class_labels_test, cluster_centers,
num_clusters=num_clusters,
distance_metric=distance_metric)
results = classifier.predict(test_vector)
elif learning_algorithm is 'yards':
yards = YARDS()
cluster_centers, classifier = yards.learn(all_class_labels_train, c=c)
test_vector, test_labels = yards.map_all(all_class_labels_test,cluster_centers)
results = classifier.predict(test_vector)
elif learning_algorithm is 'smi':
smi = SMI()
classifier = smi.learn(all_class_labels_train, c=c)
test_vector, test_labels = smi.map_all(all_class_labels_test)
results = classifier.predict(test_vector)
else:
raise Exception('Not recognized')
true_positives, true_negatives, false_positives, false_negatives = 0, 0, 0, 0
for ind, actual_labels in enumerate(test_labels):
res = int(results[ind])
if actual_labels > 0:
actual_labels = True
else:
actual_labels = False
# actual_labels = int(actual_labels)
#if type(result) is not int:
if res > 0:
res = True
else:
res = False
#results[ind] = int(results[ind])
if actual_labels and res: # true positives
true_positives = true_positives + 1
if actual_labels and not res: # false negative
false_negatives = false_negatives + 1
if not actual_labels and not res: # true negative
true_negatives = true_negatives + 1
if not actual_labels and res: # false postitives
false_positives = false_positives + 1
accuracy = (true_positives + true_negatives) / len(all_class_labels_test)
if true_positives + false_positives is 0:
precision = 0
else:
precision = true_positives / (true_positives + false_positives)
recall = true_positives / (true_positives + false_negatives)
print('accuracy: ', accuracy)
#print(accuracy)
accuracies.append(accuracy)
precisions.append(precision)
recalls.append(recall)
#print(accuracies)
#print(np.average(accuracies))
final_accuracy = np.sum(accuracies) / (bins * iterations)
final_precision = np.sum(precisions) / (bins * iterations)
final_recall = np.sum(recalls) / (bins * iterations)
#print(accuracies)
#print(precisions)
#print(recalls)
ci = 1.96 * np.std(accuracies)/math.sqrt(len(accuracies))
ci_t = ci + final_accuracy
ci_b = final_accuracy - ci
print(c, distance_metric, learning_algorithm, num_clusters)
print("Accuracy: %.3f" % final_accuracy, " %.3f" % np.std(accuracies))
print("Precision: %.3f" % final_precision, " %.3f" % np.std(precisions))
print("Recall: %.3f" % final_recall, " %.3f" % np.std(recalls))
print("CI: %.4f " %ci, " %.4f " % ci_b , "%.4f" % ci_t)
print()
def select_subset(self, bags, part_class_one, part_class_two):
"""
Categorize each example based on class label
:param examples: example datasets
:param part_class_one: examples with positive class label
:param part_class_two: examples with negative class label
:return:
"""
new_bags = []
for i in range(0, len(part_class_one)):
new_bags.append(bags[part_class_one[i]])
for i in range(0, len(part_class_two)):
new_bags.append(bags[part_class_two[i]])
return new_bags
def tune_c(c):
es = embeddedSpace()
es.ES_MI2('elephant', 0, 'dbow', distance_metric='similarity', clusters=512, C=c)
def pooled_C_tuning():
c_c = [.5, 1,10,100]
pool = multiprocessing.Pool()
pool.map(tune_c, [c for c in c_c])
pool.close()
def main():
# _,labels,data,l2,indexes = dp.loadelephant()
# bags = dp.parse_svm(data, labels, indexes)
es = embeddedSpace()
es.ES_MI('musk1', 0, 'dbow', distance_metric='euclidean', C=10, clusters=128)
if __name__ == "__main__":
multiprocessing.freeze_support()
main()