Mercurial > ift6266

annotate deep/amt/amt.py @ 403:a11692910312

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Undoing some unwanted changes
author humel
date Wed, 28 Apr 2010 11:30:37 -0400
parents 83413ac10913
children 6478eef4f8aa
rev   line source
399
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
1 # Script usage : python amt.py filname.cvs
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
2
402
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
3 import csv,numpy,re,decimal
399
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
4
402
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
5 DATASET_PATH = { 'nist' : '/data/lisa/data/ift6266h10/amt_data/nist/',
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
6 'p07' : '/data/lisa/data/ift6266h10/amt_data/p07/',
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
7 'pnist' : '/data/lisa/data/ift6266h10/amt_data/pnist/' }
399
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
8
402
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
9 CVSFILE = None
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
10 #PATH = None
403
a11692910312 Undoing some unwanted changes
humel
parents: 402
diff changeset
11 answer_labels = [ 'Answer.c'+str(i+1) for i in range(10) ]
a11692910312 Undoing some unwanted changes
humel
parents: 402
diff changeset
12 img_url = 'Input.image_url'
399
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
13 turks_per_batch = 3
402
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
14 image_per_batch = 10
399
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
15
401
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
16 def setup_association():
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
17 answer_assoc = {}
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
18 for i in range(0,10):
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
19 answer_assoc[str(i)]=i
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
20 for i in range(10,36):
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
21 answer_assoc[chr(i+55)]=i
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
22 for i in range(36,62):
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
23 answer_assoc[chr(i+61)]=i
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
24 return answer_assoc
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
25
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
26 answer_assoc = setup_association()
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
27
399
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
28 def test_error():
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
29 turks = []
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
30 reader = csv.DictReader(open(CVSFILE), delimiter=',')
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
31 entries = [ turk for turk in reader ]
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
32
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
33 errors = numpy.zeros((len(entries),))
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
34 if len(entries) % turks_per_batch != 0 :
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
35 raise Exception('Wrong number of entries or turks_per_batch')
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
36
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
37 total_uniq_entries = len(entries) / turks_per_batch
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
38
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
39 errors = numpy.zeros((len(entries),))
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
40 error_means = numpy.zeros((total_uniq_entries,))
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
41 error_variances = numpy.zeros((total_uniq_entries,))
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
42
402
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
43 PATH = DATASET_PATH[find_dataset(entries[0])]
399
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
44 for i in range(total_uniq_entries):
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
45 for t in range(turks_per_batch):
402
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
46 errors[i*turks_per_batch+t] = get_error(entries[i*turks_per_batch+t],PATH)
399
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
47 error_means[i] = errors[i*turks_per_batch:(i+1)*turks_per_batch].mean()
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
48 error_variances[i] = errors[i*turks_per_batch:(i+1)*turks_per_batch].var()
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
49
402
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
50 decimal.getcontext().prec = 3
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
51 print 'Total entries : ' + str(len(entries))
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
52 print 'Turks per batch : ' + str(turks_per_batch)
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
53 print 'Average test error : ' + str(error_means.mean()*image_per_batch) +'%'
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
54 print 'Error variance : ' + str(error_variances.mean()*image_per_batch) +'%'
399
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
55
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
56
402
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
57 def find_dataset(entry):
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
58 file = parse_filename(entry[img_url])
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
59 return file.split('_')[0]
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
60
83413ac10913 Added more stats printing. Now you dont need to parameters which dataset you are testing, it will detect it automatically
humel
parents: 401
diff changeset
61 def get_error(entry,PATH):
399
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
62 file = parse_filename(entry[img_url])
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
63 f = open(PATH+file,'r')
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
64 labels = re.sub("\s+", "",f.readline()).strip()[1:-2].split('.')
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
65 f.close()
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
66 test_error = 0
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
67 for i in range(len(answer_labels)):
401
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
68 answer = entry[answer_labels[i]]
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
69 if len(answer) != 0:
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
70 try:
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
71 if answer_assoc[answer] != int(labels[i]):
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
72 test_error+=1
399
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
73 except:
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
74 test_error+=1
401
86d5e583e278 Fixed class number bug
humel
parents: 399
diff changeset
75 else:
399
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
76 test_error+=1
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
77 return test_error
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
78
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
79
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
80 def parse_filename(string):
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
81 filename = string.split('/')[-1]
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
82 return filename.split('.')[0]+'.txt'
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
83
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
84 if __name__ =='__main__':
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
85 import sys
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
86 CVSFILE = sys.argv[1]
99905d9bc9dd Initial commit for calculating the test error of the AMT classifier
humel
parents:
diff changeset
87 test_error()