diffusionmap/example_rice.py at master · ed-li/diffusionmap · 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
import numpy as np
from scipy.spatial.distance import euclidean
from scipy.spatial.distance import mahalanobis
from diffusionmap import DiffusionMap

from sklearn.cluster import KMeans
from sklearn.decomposition import PCA
from sklearn.manifold import SpectralEmbedding

import matplotlib.pyplot as plt
import matplotlib.cm as cm


if __name__ == '__main__':
    # Load data
    data = np.loadtxt('rice.txt')

    # Plot luminance
    plt.imshow(data.mean(axis=1).reshape((93, 56)), cmap='gray', origin='lower')
    plt.axis('off')
    plt.savefig('rice_luminance.png', bbox_inches='tight')
    plt.show()

    # Diffusion map clustering based on Euclidean distances
    e_dm = DiffusionMap(data, kernel_params={'eps': 1e3}, neighbors=500)
    e_w, e_v = e_dm.map(3, 5)

    kmeans = KMeans(n_clusters=3, n_init=100)
    kmeans.fit(e_v)
    e_y = kmeans.predict(e_v)

    plt.imshow(e_y.reshape((93, 56)), origin='lower')
    plt.axis('off')
    plt.savefig('rice_euclidean.png', bbox_inches='tight')
    plt.show()

    # Diffusion map clustering based on Mahalanobis distances with overall covariances
    inv_cov = np.linalg.inv(np.cov(data, rowvar=False))
    def mdistance(x, y):
        return mahalanobis(x, y, VI=inv_cov)
    m_dm = DiffusionMap(data, kernel_params={'eps': 1e6, 'distance': mdistance}, neighbors=500)
    m_w, m_v = m_dm.map(3, 5)

    kmeans = KMeans(n_clusters=3, n_init=100)
    kmeans.fit(m_v)
    m_y = kmeans.predict(m_v)

    plt.imshow(m_y.reshape((93, 56)), origin='lower')
    plt.axis('off')
    plt.savefig('rice_mahalanobis.png', bbox_inches='tight')
    plt.show()

    # Diffusion map clustering based on Mahalanobis distances with local covariances
    lm_dm = DiffusionMap(data, kernel_params={'eps': 1e6}, neighbors=500)
    lm_w, lm_v = lm_dm.map(3, 5, local_mahalanobis=True, clusters=10)

    kmeans = KMeans(n_clusters=3, n_init=100)
    kmeans.fit(lm_v)
    lm_y = kmeans.predict(lm_v)
    print("completed kmeans")

    plt.imshow(lm_y.reshape((93, 56)), origin='lower')
    print("generated image")
    plt.axis('off')
    plt.savefig('rice_local_gmm_mahalanobis.png', bbox_inches='tight')
    plt.show()

    # # Diffusion map clustering based on Mahalanobis distances with local covariances with PCA preprocessing
    # pca = PCA(n_components=25)
    # pca.fit(data)
    # data_pca = pca.transform(data)
    #
    # plm_dm = DiffusionMap(data_pca, kernel_params={'eps': 1}, neighbors=500)
    # plm_w, plm_v = plm_dm.map(3, local_mahalanobis=True, clusters=25)
    # print("assigned mapping")
    #
    # kmeans = KMeans(n_clusters=3, n_init=100)
    # print("initialized kmeans")
    # kmeans.fit(plm_v)
    # print("fitted kmeans")
    # plm_y = kmeans.predict(plm_v)
    # print("completed kmeans")
    #
    # plt.imshow(plm_y.reshape((93, 56)), origin='lower')
    # print("generated image")
    # plt.savefig('rice_pca_local_gmm_mahalanobis.png')
    # plt.show()

    # # Diffusion map clustering based on Mahalanobis distances with local covariances with Laplacian eigenmaps preprocessing
    # le = SpectralEmbedding(n_components=50)
    # data_le = le.fit_transform(data)
    #
    # llm_dm = DiffusionMap(data_le, kernel_params={'eps': 1}, neighbors=500)
    # llm_w, llm_v = llm_dm.map(3, local_mahalanobis=True, clusters=25)
    #
    # kmeans = KMeans(n_clusters=3, n_init=100)
    # kmeans.fit(llm_v)
    # llm_y = kmeans.predict(llm_v)
    #
    # plt.imshow(llm_y.reshape((93, 56)), origin='lower')
    # plt.savefig('rice_laplacian_eigenmaps_local_gmm_mahalanobis.png')
    # plt.show()