-
Notifications
You must be signed in to change notification settings - Fork 0
/
do_convert_fake.py
298 lines (237 loc) · 8.94 KB
/
do_convert_fake.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
import os
import sys
import numpy as np
import operator
from collections import OrderedDict
import cPickle as pickle
import argparse
import multiprocessing as mp
from sklearn.preprocessing import MinMaxScaler
from sklearn.datasets import load_svmlight_file
from sklearn.datasets import dump_svmlight_file
import time
# def read_semfeat(fn):
# X = []
# with open(fn, 'rt') as f:
# for line in f:
# feature = {}
# pairs = line.strip().split()
# max_v = 0.0
# min_v = 1.0
# for pair in pairs:
# ix, v = pair.strip().split(':')
# v = float(v)
# feature[int(ix)] = v
# if v > max_v:
# max_v = v
# if v < min_v:
# min_v = v
# feature = OrderedDict(sorted(feature.items(), key=operator.itemgetter(1), reverse=True))
# vlist = feature.values()#[:10]
# klist = feature.keys()#[:10]
# feature = {k:(v - min_v)/(max_v - min_v) for (k,v) in zip(klist, vlist)}
# #feature = {ix:(v - min_v)/(max_v - min_v) for ix, v in feature.iteritems()}
# # sort them in ascending order of index
# feature = OrderedDict(sorted(feature.items(), key=operator.itemgetter(0)))
# X.append(feature)
# return X
def find_minmax(fn_in, ndim):
min_dict = {k:1 for k in range(ndim)}
max_dict = {k:0 for k in range(ndim)}
with open(fn_in, 'rt') as f:
for line in f:
# start = time.time()
feature = {}
pairs = line.strip().split()
for pair in pairs:
ix, v = pair.strip().split(':')
ix = int(ix)
v = float(v)
if min_dict[ix] > v:
min_dict[ix] = v
if max_dict[ix] < v:
max_dict[ix] = v
f.close()
return min_dict, max_dict
def fuck_semfeat(fn_in, fn_out, min_dict, max_dict, nnz):
fin = open(fn_in, 'rt')
fout = open(fn_out, 'wt')
coeffs = [1.0/(max_dict[i] - min_dict[i]) for i in range(len(min_dict))]
i = 0
for line in fin:
print(i)
i+=1
feature = {}
pairs = line.strip().split()
for pair in pairs:
ix, v = pair.strip().split(':')
ix = int(ix)
feature[ix] = (float(v) - min_dict[ix])*coeffs[ix]
# sort them in descending order of score
feature = OrderedDict(sorted(feature.items(), key=operator.itemgetter(1), reverse=True))
keys = feature.keys()[:nnz]
values = feature.values()[:nnz]
feature = {k:v for k, v in zip(keys, values)}
feature = OrderedDict(sorted(feature.items(), key=operator.itemgetter(0)))
fout.write('1 ')
for k, v in zip(feature.keys(), feature.values()):
fout.write('%d:%f ' % (k, v))
fout.write('\n')
fin.close()
fout.close()
def kick_semfeat(fn_in, fn_out, nnz):
fin = open(fn_in, 'rt')
fout = open(fn_out, 'wt')
i = 0
for line in fin:
print(i)
i+=1
feature = {}
pairs = line.strip().split()
for pair in pairs:
ix, v = pair.strip().split(':')
ix = int(ix)
feature[ix] = float(v)
# sort them in descending order of score
feature = OrderedDict(sorted(feature.items(), key=operator.itemgetter(1), reverse=True))
min_v = min(feature.values())
max_v = max(feature.values())
coeff = 1.0/(max_v - min_v)
keys = feature.keys()[:nnz]
values = feature.values()[:nnz]
values = [(v - min_v)*coeff for v in values]
feature = {k:v for k, v in zip(keys, values)}
feature = OrderedDict(sorted(feature.items(), key=operator.itemgetter(0)))
fout.write('1 ')
for k, v in zip(feature.keys(), feature.values()):
fout.write('%d:%f ' % (k, v))
fout.write('\n')
fin.close()
fout.close()
# def read_semfeat(fn):
# count = 0
# X = []
# with open(fn, 'rt') as f:
# for line in f:
# # start = time.time()
# feature = {}
# pairs = line.strip().split()
# for pair in pairs:
# ix, v = pair.strip().split(':')
# feature[int(ix)] = float(v)
# # sort them in ascending order of index
# feature = OrderedDict(sorted(feature.items(), key=operator.itemgetter(0)))
# X.append(feature)
# # end = time.time()
# count += 1
# print (count)
# return X
# WARNING: just use when x is a dense overfeat feature loaded by read_semfeat
# def dict2array(x):
# z = []
# for feature in x:
# z.append(np.array(feature.values()))
# return np.array(z)
# def dump_semfeat(X, fn, nnz, y=[]):
# if not y:
# y = [1]*len(X)
# with open(fn, 'wt') as f:
# for feat, label in zip(X, y):
# f.write('%d ' % (label))
# idx = feat.keys()
# v = feat.values()
# for idx, v in zip(feat.keys(), feat.values()):
# f.write('%d:%f ' % (idx, v))
# f.write('\n')
# f.close()
# def sparsify(x, nnz):
# z = np.zeros(x.shape)
# sorted_ix = np.argsort(x, axis=1)
# entries = sorted_ix[:, -nnz:]
# for i in range(entries.shape[0]):
# for j in range(entries.shape[1]):
# z[i, entries[i, j]] = x[i, entries[i, j]]
# return z
def main(*argv):
parser = argparse.ArgumentParser()
parser.add_argument('fn_train', help='the training feature file')
parser.add_argument('fout_train', help='the output training feature file')
parser.add_argument('fn_test', help='the training feature file')
parser.add_argument('fout_test', help='the output training feature file')
parser.add_argument('-n', '--nnz', help='the number of non-zeros retained', type=int, default=-1)
parser.add_argument('-m', '--mode', help='fuck or kick?', default='fuck')
args = parser.parse_args()
if args.mode == 'fuck':
print 'finding min, max...'
min_dict, max_dict = find_minmax(args.fn_train, 30000)
print 'converting {}'.format(args.fn_train)
p1 = mp.Process(target=fuck_semfeat, args=(args.fn_train, args.fout_train, min_dict, max_dict, args.nnz))
p1.start()
#fuck_semfeat(args.fn_train, args.fout_train, min_dict, max_dict, args.nnz)
print 'converting {}'.format(args.fn_test)
p2 = mp.Process(target=fuck_semfeat, args=(args.fn_test, args.fout_test, min_dict, max_dict, args.nnz))
p2.start()
p1.join()
p2.join()
elif args.mode == 'kick':
print 'converting {}'.format(args.fn_train)
p1 = mp.Process(target=kick_semfeat, args=(args.fn_train, args.fout_train, args.nnz))
p1.start()
#fuck_semfeat(args.fn_train, args.fout_train, min_dict, max_dict, args.nnz)
print 'converting {}'.format(args.fn_test)
p2 = mp.Process(target=kick_semfeat, args=(args.fn_test, args.fout_test, args.nnz))
p2.start()
p1.join()
p2.join()
#fuck_semfeat(args.fn_test, args.fout_test, min_dict, max_dict, args.nnz)
# print 'loading training data...'
# x_tn = read_semfeat(args.fn_train)
# x_tn = dict2array(x_tn)
# print 'loading test data...'
# x_tt = read_semfeat(args.fn_test)
# x_tt = dict2array(x_tt)
# print 'normalizing...'
# scaler = MinMaxScaler()
# x_tn = scaler.fit_transform(x_tn)
# x_tt = scaler.transform(x_tt)
# if args.nnz > 0:
# print 'sparsifying...'
# x_tn = sparsify(x_tn, args.nnz)
# x_tt = sparsify(x_tt, args.nnz)
# print 'dumping...'
# dump_svmlight_file(x_tn, [1]*x_tn.shape[0], args.fout_train)
# dump_svmlight_file(x_tt, [1]*x_tt.shape[0], args.fout_test)
if __name__ == '__main__':
sys.exit(main(*sys.argv))
##
# fn_test = '/home/phong/imdbs/VOC07/groups_367_sparse_10/semfeat_img_test_sparse_10_c_10_sknn_80_neg_300.fake.libsvm'
# fn_train = '/home/phong/imdbs/VOC07/groups_367_sparse_10/semfeat_img_train_sparse_10_c_10_sknn_80_neg_300.fake.libsvm'
# DIR_FEAT = '/home/phong/imdbs/VOC07/groups_367_sparse_10'
# x_tn = read_semfeat(fn_train)
# x_tt = read_semfeat(fn_test)
# dump_semfeat(x_tn, os.path.join(DIR_FEAT, 'semfeat_sparse_10_c_10_sknn_80_neg_300.train'), 10)
# dump_semfeat(x_tt, os.path.join(DIR_FEAT, 'semfeat_sparse_10_c_10_sknn_80_neg_300.test'), 10)
# ## ------------------
# fn_test = '/home/phong/imdbs/VOC07/groups_367_sparse_10/semfeat_img_test_sparse_10_c_10_sknn_80_neg_100k.fake.libsvm'
# fn_train = '/home/phong/imdbs/VOC07/groups_367_sparse_10/semfeat_img_train_sparse_10_c_10_sknn_80_neg_100k.fake.libsvm'
# DIR_FEAT = '/home/phong/imdbs/VOC07/groups_367_sparse_10'
# x_tn = read_semfeat(fn_train)
# x_tt = read_semfeat(fn_test)
# dump_semfeat(x_tn, os.path.join(DIR_FEAT, 'semfeat_sparse_10_c_10_sknn_80_neg_100k.train'), 10)
# dump_semfeat(x_tt, os.path.join(DIR_FEAT, 'semfeat_sparse_10_c_10_sknn_80_neg_100k.test'), 10)
##
# fn_test = '/home/phong/imdbs/VOC07/groups_367_sparse_5/semfeat_sparse_5_c_10_sknn_80_neg_300.fake.test'
# fn_train = '/home/phong/imdbs/VOC07/groups_367_sparse_5/semfeat_sparse_5_c_10_sknn_80_neg_300.fake.train'
# DIR_FEAT = '/home/phong/imdbs/VOC07/groups_367_sparse_5'
# x_tn = read_semfeat(fn_train)
# x_tt = read_semfeat(fn_test)
# dump_semfeat(x_tn, os.path.join(DIR_FEAT, 'semfeat_sparse_5_c_10_sknn_80_neg_300.train'), 5)
# dump_semfeat(x_tt, os.path.join(DIR_FEAT, 'semfeat_sparse_5_c_10_sknn_80_neg_300.test'), 5)
# ##
# fn_test = '/home/phong/imdbs/VOC07/groups_367_sparse_5/semfeat_sparse_5_c_10_sknn_80_neg_100k.fake.test'
# fn_train = '/home/phong/imdbs/VOC07/groups_367_sparse_5/semfeat_sparse_5_c_10_sknn_80_neg_100k.fake.train'
# DIR_FEAT = '/home/phong/imdbs/VOC07/groups_367_sparse_5'
# x_tn = read_semfeat(fn_train)
# x_tt = read_semfeat(fn_test)
# dump_semfeat(x_tn, os.path.join(DIR_FEAT, 'semfeat_sparse_5_c_10_sknn_80_neg_100k.train'), 5)
# dump_semfeat(x_tt, os.path.join(DIR_FEAT, 'semfeat_sparse_5_c_10_sknn_80_neg_100k.test'), 5)