-
Notifications
You must be signed in to change notification settings - Fork 0
/
certificate_fairness_gaussian_mixture.py
204 lines (156 loc) · 6.19 KB
/
certificate_fairness_gaussian_mixture.py
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
import json
import numpy as np
import os
import random
import matplotlib.pyplot as plt
import matplotlib as mpl
import seaborn as sns
import torch
def init_style(sns_style='whitegrid', font_size_base=16, linewdith_base=1.0, font="Times New Roman"):
sns.set_style(sns_style)
colors = sns.color_palette('muted')
mpl.rcParams["font.family"] = font
mpl.rcParams["mathtext.fontset"] = "stix"
mpl.rcParams["font.size"] = font_size_base
mpl.rcParams["grid.linewidth"] = linewdith_base / 2.0
mpl.rcParams["axes.linewidth"] = linewdith_base
mpl.rcParams['xtick.major.size'] = 4
mpl.rcParams['xtick.major.width'] = 1.
mpl.rcParams['ytick.major.size'] = 4
mpl.rcParams['ytick.major.width'] = 1.
return colors
# plot params
font_size = 20
linewidth = 1.0
colors = init_style(font_size_base=font_size, linewdith_base=linewidth, sns_style='darkgrid')
SEED = 742
RESULTS_DIR = ''
XLABEL = r'Distribution shift $L_2$-norm $\||\delta\||_2$'
YLABEL = 'JSD loss'
WRM_CERTIFICATE = 'WRM (Sinha et al., 2018)'
GRAMIAN_METHOD_LEGEND = 'Gramian Certificate'
SAVE_AS = ''
# init seed
np.random.seed(SEED)
random.seed(SEED)
def euclidean_distance_to_hellinger(l2_perturbations, sdev):
return np.sqrt(1 - np.exp(- l2_perturbations ** 2.0 / (8 * sdev ** 2)))
import torch.nn.functional as F
def custom_kl_div(prediction, target):
output_pos = target * (target.clamp(min=1e-7).log2() - prediction)
zeros = torch.zeros_like(output_pos)
output = torch.where(target > 0, output_pos, zeros)
output = torch.sum(output, dim=1)
return output
class JSDLoss(torch.nn.Module):
def __init__(self, num_classes, reduce='mean'):
super(JSDLoss, self).__init__()
self.num_classes = num_classes
self.reduce = reduce
def set_reduce(self, reduce):
self.reduce = reduce
def forward(self, predictions, labels):
preds = F.softmax(predictions, dim=1)
labels = F.one_hot(labels, self.num_classes).float()
distribs = [labels, preds]
mean_distrib = sum(distribs) / len(distribs)
mean_distrib_log = mean_distrib.clamp(1e-7, 1.0).log2()
kldivs1 = custom_kl_div(mean_distrib_log, labels)
kldivs2 = custom_kl_div(mean_distrib_log, preds)
if self.reduce == 'mean':
return 0.5 * (kldivs1.mean() + kldivs2.mean())
if self.reduce == 'none':
return 0.5 * (kldivs1 + kldivs2)
if self.reduce == 'sum':
return 0.5 * (kldivs1.sum() + kldivs2.sum())
assert False
# load data
data = np.load(os.path.join('./logs/', 'gaussian_mixture_data.npy'), allow_pickle=True)[()]
covariates, labels = data['test_data']
logits = data['test_logits']
checkpoint = os.path.join('./logs/model_gaussian_mixture')
sdev = 1.0
# compute loss
criterion = JSDLoss(num_classes=2)
loss = 0.
# distances
euclidean_distances = np.linspace(0, 1.0, 26)
hellinger_distances = euclidean_distance_to_hellinger(euclidean_distances, sdev)
ws_distances = euclidean_distances ** 2
# Run certification of ours
print(covariates)
print(labels)
N_P = [0] * 4
group_00, group_01, group_10, group_11 = [], [], [], []
for i,(x,y) in enumerate(zip(covariates,labels)):
if x[1] <= 0 and y == 0:
N_P[0] += 1
group_00.append(i)
elif x[1] <= 0 and y == 1:
N_P[1] += 1
group_01.append(i)
elif x[1] > 0 and y == 0:
N_P[2] += 1
group_10.append(i)
elif x[1] > 0 and y == 1:
N_P[3] += 1
group_11.append(i)
N = N_P[0] + N_P[1] + N_P[2] + N_P[3]
lambda_P = [1.0*N_P[0]/N, 1.0*N_P[1]/N, 1.0*N_P[2]/N, 1.0*N_P[3]/N]
logits = torch.tensor(logits).cuda()
labels = torch.tensor(labels).cuda()
loss_all = []
from tqdm import tqdm
for i in tqdm(range(len(logits))):
loss_all.append(criterion(logits[i:i+1], labels[i:i+1]).item())
loss_all = np.array(loss_all)
V1, V2 = [], []
V1.append(np.mean(loss_all[group_00]))
V1.append(np.mean(loss_all[group_01]))
V1.append(np.mean(loss_all[group_10]))
V1.append(np.mean(loss_all[group_11]))
V2.append(np.var(loss_all[group_00],ddof=1))
V2.append(np.var(loss_all[group_01],ddof=1))
V2.append(np.var(loss_all[group_10],ddof=1))
V2.append(np.var(loss_all[group_11],ddof=1))
# print(lambda_P)
# print(V1)
# print(V2)
# upper_bounds_general_shifting = fairness_upper_bound_general_shifting(
# hellinger_distances,
# lambda_P, V1, V2, 0, 0, 0.005)
# print(hellinger_distances)
# print(upper_bounds_general_shifting)
V1_upper = list(V1)
V1_lower = list(V1)
V2_upper = list(V2)
V2_lower = list(V2)
lambda_P_upper = list(lambda_P)
lambda_P_lower = list(lambda_P)
M = 1.0
CONFIDENCE_LEVEL=0.01
for i in range(len(V1_upper)):
V1_upper[i] = min(M, V1_upper[i] + M * np.sqrt(np.log(2 * 12 / CONFIDENCE_LEVEL) / (2 * N_P[i])))
for i in range(len(V1_lower)):
V1_lower[i] = min(M, V1_lower[i] + M * np.sqrt(np.log(2 * 12 / CONFIDENCE_LEVEL) / (2 * N_P[i])))
for i in range(len(V2_upper)):
V2_upper[i] = (np.sqrt(V2_upper[i]) + M * np.sqrt(2 * np.log(2 * 12 / CONFIDENCE_LEVEL) / (N_P[i] - 1))) ** 2
for i in range(len(V2_lower)):
V2_lower[i] = (np.sqrt(V2_lower[i]) - M * np.sqrt(2 * np.log(2 * 12 / CONFIDENCE_LEVEL) / (N_P[i] - 1))) ** 2
for i in range(len(lambda_P_upper)):
lambda_P_upper[i] = min(1, lambda_P_upper[i] + np.sqrt(np.log( 2 * 12 / CONFIDENCE_LEVEL) / (2 * N)))
lambda_P_lower[i] = lambda_P_lower[i] - np.sqrt(np.log( 2 * 12 / CONFIDENCE_LEVEL) / (2 * N))
print(f'N_P: {N_P}')
print(f'V1: {V1}')
print(f'V1_upper: {V1_upper}')
print(f'V2: {V2}')
print(f'V2_upper: {V2_upper}')
print(f'Lambda_P: {lambda_P}')
print(f'Lambda_P_upper: {lambda_P_upper}')
from fairness_bound_general_shifting_finite_sampling import fairness_upper_bound_general_shifting_finite_sampling
upper_bounds_general_shifting_finite_sampling = fairness_upper_bound_general_shifting_finite_sampling(hellinger_distances,
lambda_P, V1_upper, V1_lower, V2_upper, V2_lower, 0.0, 0.0, 0.005, lambda_P_lower, lambda_P_upper)
print(hellinger_distances)
print(upper_bounds_general_shifting_finite_sampling)
# Fairness bound subject to sensitive shifting
# upper_bounds_label_shifting_finite_sampling = fairness_certification_bound_label_shifting_finite_sampling(hellinger_distances,lambda_P_upper, V1_upper,gamma_k=args.gamma_k, gamma_r=args.gamma_r)