-
Notifications
You must be signed in to change notification settings - Fork 1
/
LLFM.py
406 lines (364 loc) · 20.9 KB
/
LLFM.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
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
'''
Tensorflow implementation of Localized Factorization Machines
'''
import math
import numpy as np
import tensorflow as tf
from sklearn.base import BaseEstimator, TransformerMixin
from sklearn.metrics import mean_squared_error
from sklearn.metrics import log_loss
from time import time
import argparse
import LoadData_nonsparse as DATA
from sparsify import sparse_concat, sparsify
from tensorflow.contrib.layers.python.layers import batch_norm as batch_norm
#################### Arguments ####################
def parse_args():
parser = argparse.ArgumentParser(description="Run FM.")
parser.add_argument('--path', nargs='?', default='data/',
help='Input data path.')
parser.add_argument('--dataset', nargs='?', default='frappe',
help='Choose a dataset.')
parser.add_argument('--epoch', type=int, default=1000,
help='Number of epochs.')
parser.add_argument('--pretrain', type=int, default=-1,
help='flag for pretrain. 1: initialize from pretrain; 0: randomly initialize; -1: save the model to pretrain file')
parser.add_argument('--batch_size', type=int, default=512,
help='Batch size.')
parser.add_argument('--hidden_factor', type=int, default=64,
help='Number of hidden factors.')
parser.add_argument('--anchor_points', type=int, default=2,
help='Number of anchor points')
parser.add_argument('--regularization_factor', type=float, default=0,
help='Regularizer for bilinear part.')
parser.add_argument('--keep_prob', type=float, default=0.5,
help='Keep probility (1-dropout_ratio) for the Bi-Interaction layer. 1: no dropout')
parser.add_argument('--lr', type=float, default=0.001,
help='Learning rate.')
parser.add_argument('--loss_type', nargs='?', default='square_loss',
help='Specify a loss type (square_loss or log_loss).')
parser.add_argument('--optimizer', nargs='?', default='AdamOptimizer',
help='Specify an optimizer type (AdamOptimizer, AdagradOptimizer, GradientDescentOptimizer, MomentumOptimizer).')
parser.add_argument('--verbose', type=int, default=1,
help='Show the results per X epochs (0, 1 ... any positive integer)')
parser.add_argument('--batch_norm', type=int, default=0,
help='Whether to perform batch normaization (0 or 1)')
return parser.parse_args()
class LLFM(BaseEstimator, TransformerMixin):
def __init__(self, features_M, pretrain_flag, save_file, hidden_factor, anchor_points, loss_type, epoch, batch_size,
learning_rate,
lambda_bilinear, keep,
optimizer_type, batch_norm, verbose, random_seed=2016, is_sparse=True):
"""
:param features_M: No. of features in the input data
:param pretrain_flag:
:param save_file:
:param hidden_factor:
:param anchor_points:
:param loss_type:
:param epoch:
:param batch_size:
:param learning_rate:
:param lamda_bilinear:
:param keep:
:param optimizer_type:
:param batch_norm:
:param verbose:
:param random_seed:
"""
# bind params to class
self.batch_size = batch_size
self.learning_rate = learning_rate
self.hidden_factor = hidden_factor
self.anchor_points = anchor_points
self.save_file = save_file
self.pretrain_flag = pretrain_flag
self.loss_type = loss_type
self.features_M = features_M
self.lambda_bilinear = lambda_bilinear
self.keep = keep
self.epoch = epoch
self.random_seed = random_seed
self.optimizer_type = optimizer_type
self.batch_norm = batch_norm
self.verbose = verbose
self.is_sparse = is_sparse
# performance of each epoch
self.train_rmse, self.valid_rmse, self.test_rmse = [], [], []
# init all variables in a tensorflow graph
self._init_graph()
def _init_graph(self):
'''
Init a tensorflow Graph containing: input data, variables, model, loss, optimizer
'''
self.graph = tf.Graph()
with self.graph.as_default(): # , tf.device('/cpu:0'):
# Set graph level random seed
tf.set_random_seed(self.random_seed)
# Input data.
if self.is_sparse:
self.train_features = tf.sparse_placeholder(tf.float32,
shape=[None, self.features_M]) # None * features_M
else:
self.train_features = tf.placeholder(tf.float32, shape=[None, self.features_M]) # None * features_M
self.train_labels = tf.placeholder(tf.float32, shape=[None, 1]) # None * 1
self.dropout_keep = tf.placeholder(tf.float32)
self.train_phase = tf.placeholder(tf.bool)
# Variables.
self.weights = self._initialize_weights()
# Model.
# coefficients
self.X2 = tf.matmul(tf.sparse_reduce_sum(tf.square(self.train_features), 1, keep_dims=True), tf.ones([1, self.anchor_points]))
self.Y2 = tf.matmul(tf.ones_like(self.train_labels, dtype=tf.float32),
tf.reduce_sum(tf.square(self.weights['anchor_points']), 0, keep_dims=True))
self.XY = tf.sparse_tensor_dense_matmul(self.train_features, self.weights['anchor_points'])
self.distance = self.X2 + self.Y2 - 2 * self.XY
self.distance = tf.sqrt(self.distance)
self.distance = -10 * self.distance
self.coefficient = tf.nn.softmax(self.distance) # None * A
# _________ sum_square part _____________
# get the summed up embeddings of features.
# Note: train_features must be a sparse, 0/1 matrix
# nonzero_embeddings = tf.nn.embedding_lookup(self.weights['feature_embeddings'], self.train_features)
# self.summed_features_emb = tf.reduce_sum(nonzero_embeddings, 1) # None * K
self.weights_reshape = tf.reshape(self.weights['feature_embeddings'], [self.features_M, self.hidden_factor * self.anchor_points])
if self.is_sparse:
self.summed_features_emb = tf.reshape(tf.sparse_tensor_dense_matmul(self.train_features,
self.weights_reshape), [-1, self.hidden_factor, self.anchor_points]) # None * K * A
else:
self.summed_features_emb = tf.reshape(tf.matmul(self.train_features,
self.weights_reshape), [-1, self.hidden_factor, self.anchor_points]) # None * K * A
# get the element-multiplication
self.summed_features_emb_square = tf.square(self.summed_features_emb) # None * K * A
# _________ square_sum part _____________
# self.squared_features_emb = tf.square(nonzero_embeddings)
# self.squared_sum_features_emb = tf.reduce_sum(self.squared_features_emb, 1) # None * K * A
if self.is_sparse:
self.squared_sum_features_emb = tf.reshape(tf.sparse_tensor_dense_matmul(tf.square(self.train_features),
tf.square(self.weights_reshape)), [-1, self.hidden_factor, self.anchor_points])
else:
self.squared_sum_features_emb = tf.reshape(tf.matmul(tf.square(self.train_features),
tf.square(self.weights_reshape)), [-1, self.hidden_factor, self.anchor_points])
# ________ FM __________
self.FM = 0.5 * tf.subtract(self.summed_features_emb_square, self.squared_sum_features_emb) # None * K * A
if self.batch_norm:
self.FM = self.batch_norm_layer(self.FM, train_phase=self.train_phase, scope_bn='bn_fm')
# TODO: How to dropout in a non-NN structure?
self.FM = tf.nn.dropout(self.FM, self.dropout_keep) # dropout at the FM layer
# _________out _________
self.Bilinear = tf.multiply(tf.reduce_sum(self.FM, 1), self.coefficient) # None * A
if self.is_sparse:
self.Feature_bias = tf.multiply(tf.sparse_tensor_dense_matmul(self.train_features, self.weights['feature_bias']), self.coefficient) # None * A
else:
self.Feature_bias = tf.multiply(tf.matmul(self.train_features, self.weights['feature_bias']), self.coefficient) # None * A
self.Bias = tf.multiply(tf.matmul(tf.ones_like(self.train_labels), self.weights['bias']), self.coefficient) # None * A
self.bilinear_reduce = tf.reduce_sum(self.Bilinear, 1)
self.feature_bias_reduce = tf.reduce_sum(self.Feature_bias, 1)
self.bias_reduce = tf.reduce_sum(self.Bias, 1)
print(self.bilinear_reduce.shape)
print(self.feature_bias_reduce.shape)
print(self.bias_reduce.shape)
self.out = tf.add_n([self.bilinear_reduce, self.feature_bias_reduce, self.bias_reduce]) # None * 1
self.out = self.out[:, tf.newaxis]
# Compute the loss.
if self.loss_type == 'square_loss':
if self.lambda_bilinear > 0:
self.loss = tf.nn.l2_loss(
tf.subtract(self.train_labels, self.out)) + tf.contrib.layers.l2_regularizer(
self.lambda_bilinear)(self.weights['feature_embeddings']) # regulizer
else:
self.loss = tf.nn.l2_loss(tf.subtract(self.train_labels, self.out))
elif self.loss_type == 'log_loss':
self.out = tf.sigmoid(self.out)
if self.lambda_bilinear > 0:
self.loss = tf.losses.log_loss(self.train_labels, self.out, weights=1.0, epsilon=1e-07,
scope=None) + tf.contrib.layers.l2_regularizer(
self.lambda_bilinear)(self.weights['feature_embeddings']) # regulizer
else:
self.loss = tf.losses.log_loss(self.train_labels, self.out, weights=1.0, epsilon=1e-07,
scope=None)
# Optimizer.
if self.optimizer_type == 'AdamOptimizer':
self.optimizer = tf.train.AdamOptimizer(learning_rate=self.learning_rate, beta1=0.9, beta2=0.999,
epsilon=1e-8).minimize(self.loss)
elif self.optimizer_type == 'AdagradOptimizer':
self.optimizer = tf.train.AdagradOptimizer(learning_rate=self.learning_rate,
initial_accumulator_value=1e-8).minimize(self.loss)
elif self.optimizer_type == 'GradientDescentOptimizer':
self.optimizer = tf.train.GradientDescentOptimizer(learning_rate=self.learning_rate).minimize(self.loss)
elif self.optimizer_type == 'MomentumOptimizer':
self.optimizer = tf.train.MomentumOptimizer(learning_rate=self.learning_rate, momentum=0.95).minimize(
self.loss)
# init
self.saver = tf.train.Saver()
init = tf.global_variables_initializer()
self.sess = tf.Session()
self.sess.run(init)
# number of params
total_parameters = 0
for variable in self.weights.values():
shape = variable.get_shape() # shape is an array of tf.Dimension
variable_parameters = 1
for dim in shape:
variable_parameters *= dim.value
total_parameters += variable_parameters
if self.verbose > 0:
print "#params: %d" % total_parameters
def _initialize_weights(self):
"""
feature_embeddings: interaction term, [features_M, K]
feature_bias: linear term, [features_M, 1]
bias: constant term, [1, 1]
:return:
"""
all_weights = dict()
if self.pretrain_flag > 0:
weight_saver = tf.train.import_meta_graph(self.save_file + '.meta')
pretrain_graph = tf.get_default_graph()
feature_embeddings = pretrain_graph.get_tensor_by_name('feature_embeddings:0')
feature_bias = pretrain_graph.get_tensor_by_name('feature_bias:0')
bias = pretrain_graph.get_tensor_by_name('bias:0')
with tf.Session() as sess:
weight_saver.restore(sess, self.save_file)
fe, fb, b = sess.run([feature_embeddings, feature_bias, bias])
all_weights['feature_embeddings'] = tf.Variable(fe, dtype=tf.float32)
all_weights['feature_bias'] = tf.Variable(fb, dtype=tf.float32)
all_weights['bias'] = tf.Variable(b, dtype=tf.float32)
else:
all_weights['feature_embeddings'] = tf.Variable(
tf.random_normal([self.features_M, self.hidden_factor, self.anchor_points], 0.0, 0.01),
name='feature_embeddings') # features_M * K * A
all_weights['feature_bias'] = tf.Variable(
tf.random_uniform([self.features_M, self.anchor_points], 0.0, 0.0),
name='feature_bias') # features_M * A
all_weights['bias'] = tf.Variable(tf.random_uniform([1, self.anchor_points]), name='bias') # 1 * A
all_weights['anchor_points'] = tf.Variable(tf.random_uniform([self.features_M, self.anchor_points]), name='anchor_points') # M * A
return all_weights
def batch_norm_layer(self, x, train_phase, scope_bn):
# Note: the decay parameter is tunable
bn_train = batch_norm(x, decay=0.9, center=True, scale=True, updates_collections=None,
is_training=True, reuse=None, trainable=True, scope=scope_bn)
bn_inference = batch_norm(x, decay=0.9, center=True, scale=True, updates_collections=None,
is_training=False, reuse=True, trainable=True, scope=scope_bn)
z = tf.cond(train_phase, lambda: bn_train, lambda: bn_inference)
return z
def partial_fit(self, data): # fit a batch
feed_dict = {self.train_features: data['X'], self.train_labels: data['Y'], self.dropout_keep: self.keep,
self.train_phase: True}
loss, opt = self.sess.run((self.loss, self.optimizer), feed_dict=feed_dict)
return loss
def get_random_block_from_data(self, data, batch_size): # generate a random block of training data
start_index = np.random.randint(0, data['Y'].shape[0] - batch_size)
if self.is_sparse:
return {
'X': sparse_concat(data['X_sparse_list'][start_index:start_index + batch_size], self.features_M),
'Y': data['Y'][start_index:start_index + batch_size, np.newaxis]
}
else:
return {
'X': data['X'][start_index:start_index + batch_size, :],
'Y': data['Y'][start_index:start_index + batch_size, np.newaxis]
}
def train(self, Train_data, Validation_data, Test_data): # fit a dataset
# Check Init performance
if self.verbose > 0:
t2 = time()
init_train = self.evaluate(Train_data)
init_valid = self.evaluate(Validation_data)
init_test = self.evaluate(Test_data)
print("Init: \t train=%.4f, validation=%.4f, test=%.4f [%.1f s]" % (
init_train, init_valid, init_test, time() - t2))
for epoch in xrange(self.epoch):
t1 = time()
total_batch = int(len(Train_data['Y']) / self.batch_size)
for i in xrange(total_batch):
# generate a batch
batch_xs = self.get_random_block_from_data(Train_data, self.batch_size)
# Fit training
self.partial_fit(batch_xs)
t2 = time()
# output validation
train_result = self.evaluate(Train_data)
valid_result = self.evaluate(Validation_data)
test_result = self.evaluate(Test_data)
self.train_rmse.append(train_result)
self.valid_rmse.append(valid_result)
self.test_rmse.append(test_result)
if self.verbose > 0 and epoch % self.verbose == 0:
print("Epoch %d [%.1f s]\ttrain=%.4f, validation=%.4f, test=%.4f [%.1f s]"
% (epoch + 1, t2 - t1, train_result, valid_result, test_result, time() - t2))
# if self.eva_termination(self.valid_rmse):
# break
if self.pretrain_flag < 0:
print "Save model to file as pretrain."
# self.saver.save(self.sess, self.save_file)
def eva_termination(self, valid):
if self.loss_type == 'square_loss':
if len(valid) > 5:
if valid[-1] > valid[-2] and valid[-2] > valid[-3] and valid[-3] > valid[-4] and valid[-4] > valid[-5]:
return True
else:
if len(valid) > 5:
if valid[-1] < valid[-2] and valid[-2] < valid[-3] and valid[-3] < valid[-4] and valid[-4] < valid[-5]:
return True
return False
def evaluate(self, data): # evaluate the results for an input set
num_example = data['Y'].shape[0]
if self.is_sparse:
feed_dict = {self.train_features: data['X_sparse'], self.train_labels: [[y] for y in data['Y']],
self.dropout_keep: 1.0, self.train_phase: False}
else:
feed_dict = {self.train_features: data['X'], self.train_labels: [[y] for y in data['Y']],
self.dropout_keep: 1.0, self.train_phase: False}
predictions = self.sess.run((self.out), feed_dict=feed_dict)
y_pred = np.reshape(predictions, (num_example,))
y_true = np.reshape(data['Y'], (num_example,))
if self.loss_type == 'square_loss':
predictions_bounded = np.maximum(y_pred, np.ones(num_example) * min(y_true)) # bound the lower values
predictions_bounded = np.minimum(predictions_bounded,
np.ones(num_example) * max(y_true)) # bound the higher values
RMSE = math.sqrt(mean_squared_error(y_true, predictions_bounded))
return RMSE
elif self.loss_type == 'log_loss':
logloss = log_loss(y_true, y_pred) # I haven't checked the log_loss
y_pred[y_pred > 0.499] = 1
y_pred[y_pred < 0.5] = 0
y_pred = y_pred.astype(dtype=np.int32)
y_true = y_true.astype(dtype=np.int32)
return np.sum(y_pred == y_true) / (num_example * 1.0)
if __name__ == '__main__':
# Data loading
args = parse_args()
data = DATA.LoadData(args.path, args.dataset, args.loss_type, False, True)
if 'X_sparse' not in data.Train_data:
data.Train_data['X_sparse_list'] = sparsify(data.Train_data['X'])
data.Train_data['X_sparse'] = sparse_concat(data.Train_data['X_sparse_list'], data.features_M)
if 'X_sparse' not in data.Validation_data:
data.Validation_data['X_sparse_list'] = sparsify(data.Validation_data['X'])
data.Validation_data['X_sparse'] = sparse_concat(data.Validation_data['X_sparse_list'], data.features_M)
if 'X_sparse' not in data.Test_data:
data.Test_data['X_sparse_list'] = sparsify(data.Test_data['X'])
data.Test_data['X_sparse'] = sparse_concat(data.Test_data['X_sparse_list'], data.features_M)
if args.verbose > 0:
print(
"FM: dataset=%s, factors=%d, loss_type=%s, #epoch=%d, batch=%d, lr=%.4f, lambda=%.1e, keep=%.2f, optimizer=%s, batch_norm=%d"
% (args.dataset, args.hidden_factor, args.loss_type, args.epoch, args.batch_size, args.lr,
args.regularization_factor, args.keep_prob, args.optimizer, args.batch_norm))
save_file = './pretrain/%s_%d/%s_%d' % (args.dataset, args.hidden_factor, args.dataset, args.hidden_factor)
# Training
t1 = time()
model = LLFM(data.features_M, args.pretrain, save_file, args.hidden_factor, args.anchor_points, args.loss_type,
args.epoch,
args.batch_size, args.lr, args.regularization_factor, args.keep_prob, args.optimizer, args.batch_norm,
args.verbose, True)
model.train(data.Train_data, data.Validation_data, data.Test_data)
# Find the best validation result across iterations
best_valid_score = 0
if args.loss_type == 'square_loss':
best_valid_score = min(model.valid_rmse)
elif args.loss_type == 'log_loss':
best_valid_score = max(model.valid_rmse)
best_epoch = model.valid_rmse.index(best_valid_score)
print ("Best Iter(validation)= %d\t train = %.4f, valid = %.4f, test = %.4f [%.1f s]"
% (best_epoch + 1, model.train_rmse[best_epoch], model.valid_rmse[best_epoch], model.test_rmse[best_epoch],
time() - t1))