forked from yang0110/graph-based-bandit
-
Notifications
You must be signed in to change notification settings - Fork 0
/
test_movielens_gob_beta.py
198 lines (159 loc) · 6.13 KB
/
test_movielens_gob_beta.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
195
196
197
import numpy as np
import networkx as nx
import seaborn as sns
sns.set_style("white")
import matplotlib.pyplot as plt
from sklearn.metrics.pairwise import rbf_kernel
from sklearn.preprocessing import Normalizer, MinMaxScaler
from scipy.sparse import csgraph
import scipy
import os
from sklearn import datasets
# os.chdir('/Kaige_Research/Code/graph_bandit/code/')
from linucb import LINUCB
from t_sampling import TS
from gob import GOB
from lapucb import LAPUCB
from lapucb_sim import LAPUCB_SIM
from club import CLUB
from utils import *
path='../bandit_results/simulated/'
#np.random.seed(2018)
user_num=20
item_num=300
dimension=10
pool_size=20
iteration=500
loop=1
sigma=0.01# noise
delta=0.1# high probability
alpha=0.1# regularizer
alpha_2=0.05# edge delete CLUB
beta=0.1 # exploration for CLUB, SCLUB and GOB
state=False # False for artificial dataset, True for real dataset
thres=0
# input_path='../processed_data/movielens/'
# user_feature_matrix_ori=np.load(input_path+'user_feature_matrix_100.npy')
# item_feature_matrix_ori=np.load(input_path+'item_feature_matrix_500.npy')
input_path='../processed_data/netflix/'
user_feature_matrix_ori=np.load(input_path+'user_feature_matrix_100.npy')
item_feature_matrix_ori=np.load(input_path+'item_feature_matrix_500.npy')
user_length=user_feature_matrix_ori.shape[0]
item_length=item_feature_matrix_ori.shape[0]
user_list=np.random.choice(range(user_length), size=20)
item_list=np.random.choice(range(item_length), size=300)
user_feature_matrix=user_feature_matrix_ori[user_list]
item_feature_matrix=item_feature_matrix_ori[item_list]
noise_matrix=np.random.normal(scale=sigma, size=(user_num, item_num))
true_payoffs=np.dot(user_feature_matrix, item_feature_matrix.T)+noise_matrix
user_seq=np.random.choice(range(user_num), size=iteration)
item_pool_seq=np.random.choice(range(item_num), size=(iteration, pool_size))
true_adj=rbf_kernel(user_feature_matrix, gamma=0.5)
np.fill_diagonal(true_adj,0)
# edges=true_adj.ravel()
# plt.figure(figsize=(5,5))
# plt.hist(edges)
# plt.ylabel('Counts', fontsize=12)
# plt.xlabel('Edge weights', fontsize=12)
# plt.tight_layout()
# plt.savefig(path+'hist_edge_weights_movielens'+'.png', dpi=100)
# plt.show()
# bandwidth_list=np.linspace(0,10,20)
# smooth_list=[]
# for ind, band in enumerate(bandwidth_list):
# adj=rbf_kernel(user_feature_matrix, gamma=band)
# smooth=np.trace(np.dot(np.dot(user_feature_matrix.T, adj), user_feature_matrix))
# smooth_list.extend([smooth])
# plt.figure(figsize=(5,5))
# plt.plot(bandwidth_list, smooth_list)
# plt.xlabel('Banditwidth')
# plt.ylabel('Smoothness')
# plt.tight_layout()
# plt.show()
D=np.diag(np.sum(true_adj, axis=1))
lap=D-true_adj
true_lap=np.zeros((user_num, user_num))
for i in range(user_num):
for j in range(user_num):
if D[i,i]==0:
true_lap[i,j]=0
else:
true_lap[i,j]=-true_adj[i,j]/D[i,i]
np.fill_diagonal(true_lap, 1)
# payoffs=true_payoffs.ravel()
# plt.figure(figsize=(5,5))
# plt.hist(payoffs)
# plt.ylabel('Counts', fontsize=12)
# plt.xlabel('Payoffs', fontsize=12)
# plt.tight_layout()
# plt.savefig(path+'hist_payoffs_movielens'+'.png', dpi=100)
# plt.show()
beta_length=5
beta_list=np.linspace(0,0.25, beta_length)
linucb_regret_matrix_beta=np.zeros((beta_length, iteration))
linucb_error_matrix_beta=np.zeros((beta_length, iteration))
gob_regret_matrix_beta=np.zeros((beta_length, iteration))
gob_error_matrix_beta=np.zeros((beta_length, iteration))
for index, b in enumerate(beta_list):
linucb_regret_matrix=np.zeros((loop, iteration))
linucb_error_matrix=np.zeros((loop, iteration))
gob_regret_matrix=np.zeros((loop, iteration))
gob_error_matrix=np.zeros((loop, iteration))
for l in range(loop):
print('loop/total_loop', l, loop)
linucb_model=LINUCB(dimension, user_num, item_num, pool_size, item_feature_matrix, user_feature_matrix, true_payoffs, 0.25, delta, sigma, state)
gob_model=GOB(dimension, user_num, item_num, pool_size, item_feature_matrix, user_feature_matrix, true_payoffs, true_adj, lap, 0.25, delta, sigma, b, state)
linucb_regret, linucb_error, linucb_beta, linucb_x_norm, linucb_inst_regret, linucb_ucb, linucb_sum_x_norm, linucb_real_beta=linucb_model.run(user_seq, item_pool_seq, iteration)
gob_regret, gob_error, gob_beta, gob_x_norm, gob_ucb, gob_sum_x_norm, gob_real_beta=gob_model.run(user_seq, item_pool_seq, iteration)
linucb_regret_matrix[l], linucb_error_matrix[l]=linucb_regret, linucb_error
gob_regret_matrix[l], gob_error_matrix[l]=gob_regret, gob_error
linucb_regret=np.mean(linucb_regret_matrix, axis=0)
linucb_error=np.mean(linucb_error_matrix, axis=0)
gob_regret=np.mean(gob_regret_matrix, axis=0)
gob_error=np.mean(gob_error_matrix, axis=0)
linucb_regret_matrix_beta[index]=linucb_regret
linucb_error_matrix_beta[index]=linucb_error
gob_regret_matrix_beta[index]=gob_regret
gob_error_matrix_beta[index]=gob_error
plt.figure(figsize=(5,5))
plt.plot(linucb_regret,'-.', markevery=0.1, label='LinUCB')
plt.plot(gob_regret, '-o', color='r', markevery=0.1, label='Gob.Lin')
plt.ylabel('Cumulative Regret', fontsize=16)
plt.xlabel('Time', fontsize=16)
plt.legend(loc=2, fontsize=16)
plt.title('b=%s'%(b))
plt.tight_layout()
#plt.savefig(path+'movielens'+'.png', dpi=100)
plt.show()
plt.figure(figsize=(5,5))
p=plt.plot(linucb_regret_matrix_beta.T)
plt.legend(p, beta_list)
plt.xlabel('Time')
plt.ylabel('Cumulative Regret')
plt.title('LinUCB')
plt.show()
plt.figure(figsize=(5,5))
p=plt.plot(gob_regret_matrix_beta.T)
plt.legend(p, beta_list)
plt.xlabel('Time')
plt.ylabel('Cumulative Regret')
plt.title('Gob.Lin')
plt.show()
plt.figure(figsize=(5,5))
plt.plot(linucb_regret,'-.', markevery=0.1, label='LinUCB')
plt.plot(gob_regret, '-o', color='r', markevery=0.1, label='Gob.Lin')
plt.ylabel('Cumulative Regret', fontsize=16)
plt.xlabel('Time', fontsize=16)
plt.legend(loc=2, fontsize=16)
plt.tight_layout()
#plt.savefig(path+'movielens'+'.png', dpi=100)
plt.show()
plt.figure(figsize=(5,5))
plt.plot(linucb_error,'-.', markevery=0.1, label='LinUCB')
plt.plot(gob_error, '-*',color='r', markevery=0.1, label='Gob.Lin')
plt.ylabel('Error', fontsize=12)
plt.xlabel('Time', fontsize=12)
plt.legend(loc=1, fontsize=12)
plt.tight_layout()
#plt.savefig(path+'error'+'.png', dpi=100)
plt.show()