sapimouse/main_mouse_blocks.py at main · margitantal68/sapimouse · 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
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
import pandas as pd
import numpy as np

from util.oneclass import  evaluate_authentication_train_test
from util.model import get_model_output_features, train_model
from util.normalization  import normalize_rows, normalize_columns, normalize_all

from util.classification import evaluate_identification_Train_Test
from util.settings import DataType, RepresentationType
import util.settings as st

from util.plot import plot_ROC_filelist, plot_ROC_single
from util.classification import evaluate_identification_CV
from util.utils import create_userids

from enum import Enum

class RawDataType( Enum ):
    vx_vy = "vx_vy"
    ABS_vx_vy = "ABS_vx_vy"
    dx_dy = "dx_dy"
    ABS_dx_dy = "ABS_dx_dy"


# feature learning
# used subjects: 1..72 (session 3 min + session 1 min)
def train_feature_extractor():
    df1 = pd.read_csv("input_csv_mouse/sapimouse_" + raw_data_type.value +"_3min.csv")
    df2 = pd.read_csv("input_csv_mouse/sapimouse_" + raw_data_type.value +"_1min.csv")
    df1 = normalize_rows( df1, st.NormalizationType.ZSCORE)
    df2 = normalize_rows( df2, st.NormalizationType.ZSCORE)

    users_train = [x for x in range(1,73)]
    df1_train =  df1.loc[ df1.iloc[:, -1].isin([ x for x in users_train ]) ]
    df2_train =  df2.loc[ df2.iloc[:, -1].isin([ x for x in users_train ]) ]

    frames = [df1_train, df2_train]
    df_train = pd.concat(frames)
    train_model(df_train, model_name, num_filters, representation_learning=True)


# users_eval - subsetset of users used for evaluation (session 3 min + session 1 min)
# num_blocks - number of blocks used for score computations

def verification_block_traintest( users_eval, num_blocks = 1, verbose = False):
    df1 = pd.read_csv("input_csv_mouse/sapimouse_" + raw_data_type.value +"_3min.csv")
    df2 = pd.read_csv("input_csv_mouse/sapimouse_" + raw_data_type.value +"_1min.csv")
    df1 = normalize_rows( df1, st.NormalizationType.ZSCORE)
    df2 = normalize_rows( df2, st.NormalizationType.ZSCORE)

    df1_eval =  df1.loc[ df1.iloc[:, -1].isin(users_eval) ]
    df2_eval =  df2.loc[ df2.iloc[:, -1].isin(users_eval) ]

    # Feature extraction
    df1_eval_features = get_model_output_features( df1_eval, model_name )
    df2_eval_features = get_model_output_features( df2_eval, model_name )

    # Authentication
    roc_data_filename2 = 'results/roc_' + str(num_blocks) + '.csv'
    evaluate_authentication_train_test( df1_eval_features, df2_eval_features, data_type, num_blocks, representation_type, verbose = verbose, roc_data = True, roc_data_filename = roc_data_filename2)

    # plot ROC
    title = 'ROC curve ' + str(data_type.value) + '_' + str(raw_data_type.value)
    plot_ROC_single(roc_data_filename2, title = title)


# VERIFICATION
# input parameters:
#       num_blocks - number of blocks used for decision
# output:
#       file containing positive and negatives scores
#       results/roc_[num_blocks].csv

def verification(num_blocks):
    users = range(73,121)
    verification_block_traintest(users, num_blocks = num_blocks, verbose = False)


# MAIN
if __name__ == "__main__":
    dataset_name ='sapimouse_72_'
    data_type = DataType.MOUSE
    # Select the type of raw features
    raw_data_type = RawDataType.ABS_dx_dy
    # number of FCN filters
    num_filters = 128
    representation_type = RepresentationType.EE
    model_name = dataset_name + '_fcn_' + str(num_filters) + '_' + str(raw_data_type.value) + '.h5'

    # Train feature extractor
    # train_feature_extractor()

    # User authentication/verification
    for num_blocks in range(1,6):
        print('Number of blocks: ', num_blocks)
        verification(num_blocks)

    # ROC curves for blocks 1..5
    filelist = ['results/roc_1.csv', 'results/roc_2.csv', 'results/roc_3.csv', 'results/roc_4.csv', 'results/roc_5.csv']
    plot_ROC_filelist(filelist, title = 'Sapimouse - 48 users', outputfilename='output_png/roc_sapimouse_48.png')


# UNIT = block of 128 events
# def classification_block_traintest( users, raw_data_type = RawDataType.vx_vy ):
#     num_filters = 128
#     df1 = pd.read_csv("input_csv_mouse/sapimouse_" + raw_data_type.value +"_3min.csv")
#     df2 = pd.read_csv("input_csv_mouse/sapimouse_" + raw_data_type.value +"_1min.csv")

#     df1 =  df1.loc[ df1.iloc[:, -1].isin(users) ]
#     df2 =  df2.loc[ df2.iloc[:, -1].isin(users) ]

#     df1 = normalize_rows( df1, st.NormalizationType.ZSCORE)
#     df2 = normalize_rows( df2, st.NormalizationType.ZSCORE)

#     # EE features
#     model_name = dataset_name + '_fcn_' + str(num_filters) + '_' + str(raw_data_type.value) + '.h5'
#     # model_name = 'balabit_fcn_128.h5'
#     print(model_name)
#     df1_features = get_model_output_features( df1, model_name )
#     df2_features = get_model_output_features( df2, model_name )

#     evaluate_identification_Train_Test(df1_features, df2_features)

# # UNIT = block of 128 events
# def classification_block_train( users, raw_data_type = RawDataType.vx_vy ):
#     num_filters = 128
#     df1 = pd.read_csv("input_csv_mouse/sapimouse_" + raw_data_type.value +"_3min.csv")
#     df1 =  df1.loc[ df1.iloc[:, -1].isin(users) ]

#     df1 = normalize_rows( df1, st.NormalizationType.ZSCORE)

#     # EE features
#     model_name = dataset_name + '_fcn_' + str(num_filters) + '_' + str(raw_data_type.value) + '.h5'
#     # model_name = 'balabit_fcn_128.h5'
#     df1_features = get_model_output_features( df1, model_name )
#     evaluate_identification_CV(df1_features)


# # UNIT = block of 128 events
# # classification based on raw data, without the feature learning step
# def classification_block_raw_traintest( users, raw_data_type = RawDataType.vx_vy ):
#     num_filters = 128
#     df1 = pd.read_csv("input_csv_mouse/sapimouse_" + raw_data_type.value +"_3min.csv")
#     df1 =  df1.loc[ df1.iloc[:, -1].isin(users) ]
#     df1 = normalize_rows( df1, st.NormalizationType.ZSCORE)

#     df2 = pd.read_csv("input_csv_mouse/sapimouse_" + raw_data_type.value +"_1min.csv")
#     df2 =  df2.loc[ df2.iloc[:, -1].isin(users) ]
#     df2 = normalize_rows( df2, st.NormalizationType.ZSCORE)
#     evaluate_identification_Train_Test(df1, df2)


# # CLASSIFICATION/IDENTIFICATION
# # feature learning: sapimouse users 1 .. 72
# # classification: Random Forests (100 trees)
# #                 training: 3 min session
# #                 test: 1 min session
# def classification(raw_data_type):
#     users = range(72,121)
#     classification_block_traintest( users, raw_data_type= raw_data_type )