data-augmented-metamorphic-testing/my_functions.py at master · aravi11/data-augmented-metamorphic-testing · 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
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
import numpy as np
import pandas as pd
import networkx
from scipy.sparse import lil_matrix, kron,identity
from scipy.sparse.linalg import lsqr
from sklearn.utils import resample

def norm_mat(adj_mat):
    norm = adj_mat.sum(axis=0)
    norm[norm == 0] = 1
    return adj_mat / norm

def my_kernel(X, Y, lmb):

    max_size = int(np.sqrt(X.shape[1]))
    RWK = np.zeros([X.shape[0], Y.shape[0]])
    step = 10

    for i in range(0, X.shape[0]):
        for j in range(0, Y.shape[0]):
            weighted_sum = 0
            am_pg = np.kron(norm_mat(np.reshape(X[i, :], (max_size, max_size))), norm_mat(np.reshape(Y[j, :], (max_size, max_size))))
            for k in range(step):
                weighted_sum += np.dot(lmb ** k, am_pg ** k)
            rwk = weighted_sum.sum()
            RWK[i, j] = rwk

    return RWK

def random_walk_kernel(X, Y, lmb):

    max_size = int(np.sqrt(X.shape[1]))
    kernel_matrix = np.zeros([X.shape[0], Y.shape[0]])
    step = 10

    for i in range(0, X.shape[0]):
        g1 = lil_matrix(norm_mat(np.reshape(X[i, :], (max_size, max_size))))
        for j in range(0, Y.shape[0]):
            g2 = lil_matrix(norm_mat(np.reshape(Y[j, :], (max_size, max_size))))
            weighted_sum = 0
            am_pg = kron(g1, g2)
            for k in range(step):
                weighted_sum += (am_pg ** k).dot(lmb ** k)
            kernel_matrix[i, j] = weighted_sum.sum()

    return kernel_matrix

def random_walk_kernel_1(X, Y, lmb):

    max_size = int(np.sqrt(X.shape[1]))
    kernel_matrix = np.zeros([X.shape[0], Y.shape[0]])

    for i in range(0, X.shape[0]):
        g1 = norm_mat(np.reshape(X[i, :], (max_size, max_size)))
        for j in range(0, Y.shape[0]):
            g2 = norm_mat(np.reshape(Y[j, :], (max_size, max_size)))
            w_prod = kron(lil_matrix(g1), lil_matrix(g2))
            starting_prob = np.ones(w_prod.shape[0]) / (w_prod.shape[0])
            stop_prob = starting_prob
            A = identity(w_prod.shape[0]) - (w_prod * lmb)
            x = lsqr(A, starting_prob)
            kernel_matrix[i, j] = stop_prob.T.dot(x[0])

    return kernel_matrix

def compute_kernel_matrix(X, Y, lmb, type):

    kernel_matrix = np.zeros([len(X), len(Y)])

    for i in range(0, len(X)):
        for j in range(0, len(Y)):
            if type == "RWK":
                kernel_matrix[i, j] = RWK(X[i], Y[j], lmb)
            elif type == "RWK_norm":
                kernel_matrix[i, j] = RWK_norm(X[i], Y[j], lmb)
            elif type == "RWK_1":
                kernel_matrix[i, j] = RWK_1(X[i], Y[j], lmb)
            elif type == "RWK_1_norm":
                kernel_matrix[i, j] = RWK_1_norm(X[i], Y[j], lmb)

    return kernel_matrix

def RWK(X, Y, lmb):

    step = 10
    weighted_sum = 0
    g1 = norm_mat(networkx.adjacency_matrix(X))
    g2 = norm_mat(networkx.adjacency_matrix(Y))
    g_prod = kron(lil_matrix(g1), lil_matrix(g2))

    for n in range(step):
        weighted_sum += (g_prod ** n).dot(lmb ** n)

    k = weighted_sum.sum()

    return k

def RWK_norm(X, Y, lmb):

    step = 10
    weighted_sum = 0
    weighted_sum_1 = 0
    weighted_sum_2 = 0
    g1 = norm_mat(networkx.adjacency_matrix(X))
    g2 = norm_mat(networkx.adjacency_matrix(Y))
    g_prod = kron(lil_matrix(g1), lil_matrix(g2))
    g_prod_1 = kron(lil_matrix(g1), lil_matrix(g1))
    g_prod_2 = kron(lil_matrix(g2), lil_matrix(g2))

    for n in range(step):
        weighted_sum += (g_prod ** n).dot(lmb ** n)
        weighted_sum_1 += (g_prod_1 ** n).dot(lmb ** n)
        weighted_sum_2 += (g_prod_2 ** n).dot(lmb ** n)

    k = weighted_sum.sum()
    k_1 = weighted_sum_1.sum()
    k_2 = weighted_sum_2.sum()
    k_norm = k / np.sqrt(k_1 * k_2)

    return k_norm

def RWK_1(X, Y, lmb):

    g1 = norm_mat(networkx.adjacency_matrix(X))
    g2 = norm_mat(networkx.adjacency_matrix(Y))
    g_prod = kron(lil_matrix(g1), lil_matrix(g2))
    starting_prob = np.ones(g_prod.shape[0]) / (g_prod.shape[0])
    stop_prob = starting_prob
    A = identity(g_prod.shape[0]) - (g_prod * lmb)
    x = lsqr(A, starting_prob)
    k = stop_prob.T.dot(x[0])

    return k

def RWK_1_norm(X, Y, lmb):

    g1 = norm_mat(networkx.adjacency_matrix(X))
    g2 = norm_mat(networkx.adjacency_matrix(Y))
    g_prod = kron(lil_matrix(g1), lil_matrix(g2))
    g_prod_1 = kron(lil_matrix(g1), lil_matrix(g1))
    g_prod_2 = kron(lil_matrix(g2), lil_matrix(g2))
    starting_prob = np.ones(g_prod.shape[0]) / (g_prod.shape[0])
    starting_prob_1 = np.ones(g_prod_1.shape[0]) / (g_prod_1.shape[0])
    starting_prob_2 = np.ones(g_prod_2.shape[0]) / (g_prod_2.shape[0])
    stop_prob = starting_prob
    stop_prob_1 = starting_prob_1
    stop_prob_2 = starting_prob_2
    A = identity(g_prod.shape[0]) - (g_prod * lmb)
    A_1 = identity(g_prod_1.shape[0]) - (g_prod_1 * lmb)
    A_2 = identity(g_prod_2.shape[0]) - (g_prod_2 * lmb)
    x = lsqr(A, starting_prob)
    x_1 = lsqr(A_1, starting_prob_1)
    x_2 = lsqr(A_2, starting_prob_2)
    k = stop_prob.T.dot(x[0])
    k_1 = stop_prob_1.T.dot(x_1[0])
    k_2 = stop_prob_2.T.dot(x_2[0])
    k_norm = k / np.sqrt(k_1 * k_2)

    return k_norm

def balance_data(G, label, random_state):
    graphs = {'CFG': G, 'label': label}
    #print(graphs)
    df = pd.DataFrame(data=graphs)

    if len(df[df.label == -1]) > len(df[df.label == 1]):
        df_majority = df[df.label == -1]
        df_minority = df[df.label == 1]
    else:
        df_majority = df[df.label == 1]
        df_minority = df[df.label == -1]

    df_minority_upsampled = resample(df_minority, replace=True, n_samples=len(df_majority), random_state=random_state)
    df_upsampled = pd.concat([df_majority, df_minority_upsampled], ignore_index=True)

    data = np.asarray(df['CFG'])
    target = np.asarray(df['label'])

    return data, target

def rename_nodes(node_label):

    n = len(node_label)
    print(node_label)
    labels = [0] * n
    label_lookup = {}
    label_counter = 0

    for i in range(n):
        num_nodes = len(node_label[i])
        temp_lables = list(node_label[i])
        labels[i] = np.zeros(num_nodes, dtype=np.uint64)
        for j in range(num_nodes):
            temp_node_str = str(np.copy(temp_lables[j]))
            if temp_node_str not in label_lookup:
                label_lookup[temp_node_str] = label_counter
                labels[i][j] = label_counter
                label_counter += 1
            else:
                labels[i][j] = label_lookup[temp_node_str]

    L = label_counter
    print('Number of original labels: %d' % L)

    return L, labels