anomaly-explain/scramble.py at master · lgilpin/anomaly-explain · GitHub

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
import logging
import numpy as np
import argparse

"""
Perform the binomial distribution (return a 0 or 1)
"""
def coinFlip(p):
    result = np.random.binomial(1,p)
    return result

def flip():
    return coinFlip(.5)

def plusMinus():
    if flip()==1:
        return 1
    else: return -1

def scrambleFlags(n, p):
    flags = np.arange(n)
    #perform desired numbered of flips at required probability set above
    for i in range(0, n):
        flags[i] = coinFlip(p)
        i+=1

    #Results
    logging.debug('probability is set to {:f}'.format(p))
    logging.debug(flags)

    reasonable_count = np.count_nonzero(flags == 1)
    unreasonable_count = np.count_nonzero(flags == 0)

    #Total up heads and tails for easy user experience
    logging.debug('Reasonable Count: {:d}'.format(reasonable_count))
    logging.debug('Uneasonable Count: {:d}'.format(unreasonable_count))

    logging.debug('Reasonable Percentage: {:f}'.format(reasonable_count/n))
    logging.debug('Unreasonable Percentage: {:f}'.format(unreasonable_count/n))

    return flags

def chooseNewLabel(old_label, labels, probs):
    """
    Chooses a new label based on a distribution multi-class distribution
    labels are a dictionary of the label name and the probability
    """
    choice = np.random.choice(labels, p=probs)
    while choice == old_label:
        choice = np.random.choice(labels, p=probs)
    return choice

def toggleSizes(size, std, debug=False):
    """"
    Add or subtract noise from a bounding box
    """
    toggled = [0,0,0]
    for i in range(0,2):
        toggled[i]= size[i]+plusMinus()*np.random.uniform(0,2*std[i])
    logging.debug("Originally", size, " changed to ", toggled)
    return toggled

def add_noise(std):
    noise = np.random.uniform(0,2*std)
    if flip(): # 0 or 1 with 50% probability
        return noise
    else: return -noise

def printDistribution(p, flags):
    n = len(flags)

    reasonable_count = np.count_nonzero(flags == 1)
    unreasonable_count = np.count_nonzero(flags == 0)

    #Total up heads and tails for easy user experience
    print('Reasonable Count: {:d}'.format(reasonable_count))
    print('Uneasonable Count: {:d}'.format(unreasonable_count))

    print('Reasonable Percentage: {:f}'.format(reasonable_count/n))
    print('Unreasonable Percentage: {:f}'.format(unreasonable_count/n))


def main():
    """
    Verbose shows the dataset distributions etc.
    """
    parser = argparse.ArgumentParser()
    parser.add_argument("-v", "--verbose", action='store_true',
                        help='This is the same as debug right now')
    parser.add_argument("-n", "--number", help="The number of items toscramble",
                        type=int, default=10)
    parser.add_argument("-p", "--probability",
                        help="The probability for the binomial distribution",
                        type=float, default=.5)
    args = parser.parse_args()

    if args.verbose:
        logging.basicConfig(level=logging.DEBUG,
                            format='%(levelname)s: %(message)s')

    # Logging info for verbose and other debugging
    logging.info("Verbose output.")

    scrambleFlags(args.number, args.probability)

if __name__ == "__main__":
    # execute only if run as a script
    main()