VisGraphsOnGraphs/detection.py at main · koc-lab/VisGraphsOnGraphs · 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
import numpy as np
import argparse
from pygsp import graphs
import scipy.sparse as sp
from gvg import GVG

def GraphStat(G_v):
    deg = np.asarray(G_v.sum(axis=1)).reshape(-1)
    return float(deg.mean())


def H0_Sampler(W, x, args):
    R = args.R
    alpha = args.alpha
    trial_n = args.trial_n
    logging = args.logging
    mode = args.mode
    multi = args.multi

    N = W.shape[0]
    T0 = np.empty(trial_n, dtype=float)
    gvg = GVG(W, R=R, mode=mode, logging=logging)

    for trial in range(trial_n):
        np.random.seed(trial)
        x_p = np.random.permutation(x)
        if multi:
            G_vp, dist_p = gvg.build_parallel(x_p)
        else:
            G_vp, dist_p = gvg.build(x_p)
        T0[trial] = GraphStat(G_vp)

    # H0 stat and threshold
    W_H0 = np.mean(T0)
    tau = np.quantile(T0, q=1-alpha)
    return tau, T0

def GVG_Detector(W, x, tau, args):
    R = args.R
    alpha = args.alpha
    trial_n = args.trial_n
    logging = args.logging
    mode = args.mode
    multi = args.multi

    gvg = GVG(W, R=R, mode=mode, logging=logging)

    # Now H1 stat
    if multi:
        G_v, dist = gvg.build_parallel(x)
    else:
        G_v, dist = gvg.build(x)

    T_obs = GraphStat(G_v)
    return T_obs, (T_obs >= tau)


if __name__ == "__main__":

    parser = argparse.ArgumentParser()
    parser.add_argument("--R", type=int, default=5)
    parser.add_argument("--alpha", type=float, default=0.01)
    parser.add_argument("--trial_n", type=int, default=10)
    parser.add_argument("--logging", action="store_true")
    parser.add_argument("--mode", type=str, default="all")
    parser.add_argument("--multi", action="store_true")
    args = parser.parse_args()

    np.random.seed(0)

    N = 10000
    G = graphs.Sensor(N, k=5)

    W = G.W.tocsr()
    W = (W > 0).astype(np.int8).tocsr()
    print("Num nodes:", G.N)
    print("Num edges:", W.nnz)

    v0 = np.random.randint(0, N)

    from scipy.sparse.csgraph import dijkstra
    dist = dijkstra(W, directed=False, unweighted=True, indices=v0)
    dist[~np.isfinite(dist)] = np.max(dist[np.isfinite(dist)]) + 1

    p = 2
    g = (dist / dist.max())**p

    A = 20.0
    x = np.random.randn(N)
    x += A * g


    tau, T0 = H0_Sampler(W, x, args)
    T_obs, is_H1 = GVG_Detector(W, x, tau, args)
    print(T_obs)
    print(tau)
    print(is_H1)
    print(T0)

    args.multi = True
    tau, T0 = H0_Sampler(W, x, args)
    T_obs, is_H1 = GVG_Detector(W, x, tau, args)
    print(T_obs)
    print(tau)
    print(is_H1)
    print(T0)