IML_for_Imbalanced_Learning/k_mediod_viz.py at main · dd1github/IML_for_Imbalanced_Learning · 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
# -*- coding: utf-8 -*-

from sklearn_extra.cluster import KMedoids
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from imbalance_cifar import IMBALANCECIFAR10


#inputs
#the inputs are based on CIFAR-10 as an illustration
###################################################################

#number of classes
num_class = 10

#number of archetypes
num_arch = 4

#file location with source data
#this file is created with arch_NNB.py
source_file = ".../CE_cif_trn_NNB.csv"

#source file with instance embeddings (FE)
#file created with Extract_FE.py
source_FE = ".../CE_cifar10_train.csv"

#folder to save the prototypical images
save_folder = ".../protos/"

# class target index labels
select = np.array([0,9])

#class labels of interest
classes = ['Plane','Truck']

#number of classes selected
len_cls_select = len(classes)

#archetype labels
arch = ['Safe','Border','Rare','Noise']

#location where image dataset stored
data_root = '.../data/'

#######################################################################

train_dataset = IMBALANCECIFAR10(root=data_root,
        imb_type='exp', imb_factor=0.01,
        rand_number=0, train=True, download=True)


pd_ce = pd.read_csv(source_file)
x = pd_ce.to_numpy()

#extract targets
tar = x[:,0]
#extract indices
idx = x[:,1]
#extract CNN class predictions
pred = x[:,2]
#extract categories or archetypes
cat = x[:,3]


pdf = pd.read_csv(source_FE)
x1 = pdf.to_numpy()

#extract FE
xfeat = x1[:,3:]

medoids = np.empty([len_cls_select,num_arch],dtype=int)

count = 0
for i in select:
    tars = tar[tar==i]
    idxs = idx[tar==i]
    cats = cat[tar==i]
    feats = xfeat[tar==i]

    for c in range(num_arch):
        ctars = tars[cats==c]
        cfeats = feats[cats==c]
        cidx = idxs[cats==c]
        kmedoids = KMedoids(n_clusters=1, random_state=0).fit(cfeats)
        med = kmedoids.medoid_indices_
        medoids[count,c]=cidx[med]

    count+=1


for i in range(len(select)):
    for j in range(num_arch):

        img_idx = medoids[i,j]

        img_idx = img_idx.astype(int)

        plt.imshow(train_dataset.data[img_idx])
        plt.title(classes[i])

        plt.axis('off')

        f3 = save_folder + classes[i] +'_' + arch[j] + '.pdf'

        plt.savefig(f3,bbox_inches='tight')
        plt.show()