TPAMTL/visualize_final_total_custom.py at master · mandiehyewon/TPAMTL · 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
import matplotlib.pyplot as plt
import numpy as np
import pdb
import pickle
import scipy
import os
import seaborn as sns

#dataset = 'mimic-infection'
#path = '/st2/tuan_home/AMTL/images_mimic/'

dataset = 'physionet'
path = './images_physionet/'

if dataset == 'physionet':
    num_steps = 49
    num_features = 8
    num_samples = 30
    num_tasks = 4
    task = ['Mortality','Stay<3','Cardiac','Recovery']

elif dataset == 'mimic-infection':
    num_steps = 48
    num_features = 8
    num_samples = 30
    num_tasks = 4
    task = ['Fever','Leukocytosis','Infection','Mortality']

beta_outputs = pickle.load(open(path+'beta_outputs.pkl','rb'))
att_eachs = pickle.load(open(path+'att_eachs.pkl','rb'))

beta_outputs_task = [None for _ in range(4)]

for i in range(4):
    beta_outputs_task[i] = [x[i] for x in beta_outputs]


pid = 1


folder = path + 'customm/%d/'%(pid)

# amount-source
source_id = 2
target_id = 0
steps = range(0,23)
w_i = 0
w_s = None

source_avr_vars = []
source_avr_means = []
for step in steps:
    total_var = 0
    total_mean = 0
    for f in range(num_features):
        total_var += np.var([beta_outputs_task[source_id][sam][pid][step][f] for sam in range(num_samples)])
        total_mean += np.mean([beta_outputs_task[source_id][sam][pid][step][f] for sam in range(num_samples)])
    avr_var = total_var / num_features
    avr_mean = total_mean / num_features
    source_avr_vars.append(avr_var)
    source_avr_means.append(avr_mean)

transfer_total = []
for step in steps:
    transfers = []
    transfer_stds = []
    target_step_max = num_steps if w_s==None else min(num_steps,step+w_s)
    for target_step in range(step,target_step_max):
        transfer = np.mean([att_each[target_id,source_id][w_i][pid][target_step,step] for att_each in att_eachs])
        transfers.append(transfer)
    transfers = np.mean(transfers)
    transfer_total.append(transfers)


fig,ax = plt.subplots(figsize=[10,3.5])
color = 'red'
ax = sns.lineplot(steps, source_avr_vars,marker='.',color=color,ax=ax)
ax.legend(loc='upper left')
ax.set_ylabel('Uncertainty of Source',fontsize=18)
ax.set_ylim(top=max(source_avr_vars)*1.1)
ax.margins(y=0)
ax.spines['right'].set_color(color)
ax.tick_params(axis='y', colors=color)
ax.yaxis.label.set_color(color)


ax1 = ax.twinx()
sns.lineplot(steps, transfer_total,marker='.', color='#2976b1',ax=ax1)
ma = max(transfer_total)
mi = min(transfer_total)
ax1.set_ylim([mi,ma])
ax1.set_ylabel('Transfer Amount',fontsize=18)
ax1.spines['right'].set_color('#2976b1')
ax1.tick_params(axis='y', colors='#2976b1')
ax1.yaxis.label.set_color('#2976b1')
# ax1.set_ylim([mi*3/4,ma*2])
ax1.set_xticks(steps)
ax1.legend(loc="upper right")
ax1.margins(y=0)
ax.set_title('Knowledge Transfer from %s to %s: \nUncertainty of Source features vs Normalized Total Knowledge transfer'%(task[source_id],task[target_id]),fontsize=18)

if not os.path.exists(folder+'/Transfer%s_%s/'%(task[source_id],task[target_id])):
    os.makedirs(folder+'/Transfer%s_%s/'%(task[source_id],task[target_id]))
ax = fig.get_figure()
fig.savefig(folder+'/Transfer%s_%s/source_amount.png'%(task[source_id],task[target_id]),bbox_inches='tight')
plt.close(fig)


# amount-target
source_id = 0
target_id = 2
steps = range(15,38)
w_i = 0
w_s = None

transfer_total = []
for step in steps:
    transfers = []
    source_step_min = 0 if w_s==None else max(0,step-w_s+1)
    for source_step in range(source_step_min,step+1):
        transfer = np.mean([att_each[target_id,source_id][w_i][pid][step,source_step] for att_each in att_eachs])
        transfers.append(transfer)
    transfers = np.mean(transfers)
    transfer_total.append(transfers)

target_avr_vars = []
target_avr_means = []

for step in steps:
    total_var = 0
    total_mean = 0
    for f in range(num_features):
        total_var += np.var([beta_outputs_task[target_id][sam][pid][step][f] for sam in range(num_samples)])
        total_mean += np.square(np.mean([beta_outputs_task[target_id][sam][pid][step][f] for sam in range(num_samples)]))
    avr_var = total_var / num_features
    avr_mean = np.sqrt(total_mean / num_features)
    target_avr_vars.append(avr_var)
    target_avr_means.append(avr_mean)

fig,ax = plt.subplots(figsize=[10,3.5])
color = 'red'
ax = sns.lineplot(steps, target_avr_vars,marker='.',color=color,ax=ax)
ax.set_ylim(top=max(target_avr_vars)*1.1)
ax.set_ylabel('Uncertainty of Target',fontsize=18)
ax.legend(loc='upper left')

ax.spines['right'].set_color(color)
ax.tick_params(axis='y', colors=color)
ax.yaxis.label.set_color(color)

ax1 = ax.twinx()
sns.lineplot(steps, transfer_total,marker='.', color='#2976b1',ax=ax1)
ma = max(transfer_total)
mi = min(transfer_total)
ax1.set_ylim(mi,ma)
ax1.set_ylabel('Transfer Amount',fontsize=18)
ax1.spines['right'].set_color('#2976b1')
ax1.tick_params(axis='y', colors='#2976b1')
ax1.yaxis.label.set_color('#2976b1')
# ax1.set_ylim([mi*3/4,ma*2])
ax1.set_xticks(steps)
ax1.legend(loc="upper right")
ax.set_title('Knowledge Transfer from %s to %s: \nUncertainty of Target Features vs Normalized Total Knowledge transfer'%(task[source_id],task[target_id]),fontsize=18)

if not os.path.exists(folder+'/Transfer%s_%s/'%(task[source_id],task[target_id])):
    os.makedirs(folder+'/Transfer%s_%s/'%(task[source_id],task[target_id]))
fig = ax.get_figure()
fig.savefig(folder+'/Transfer%s_%s/target_amount.png'%(task[source_id],task[target_id]),bbox_inches='tight')
plt.close(fig)