forked from XifengGuo/DEC-keras
-
Notifications
You must be signed in to change notification settings - Fork 0
/
DEC.py
244 lines (198 loc) · 9.12 KB
/
DEC.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
"""
Keras implementation for Deep Embedded Clustering (DEC) algorithm:
Junyuan Xie, Ross Girshick, and Ali Farhadi. Unsupervised deep embedding for clustering analysis. ICML 2016.
Usage:
use `python DEC.py -h` for help.
Author:
Xifeng Guo. 2017.1.30
"""
from time import time
import numpy as np
import tensorflow.keras.backend as K
from tensorflow.python.keras.optimizers import SGD
from tensorflow.python.keras.engine import Layer, InputSpec
from tensorflow.python.keras.models import Model
from sklearn.cluster import KMeans
import metrics
import tensorflow as tf
import tensorflow_hub as hub
universal = hub.Module("https://tfhub.dev/google/universal-sentence-encoder/2")
def UniversalEmbedding(sen):
return universal(tf.squeeze(tf.cast(sen, tf.string)))
def autoencoder(dims, act='relu', init='glorot_uniform'):
"""
Fully connected auto-encoder model, symmetric.
Arguments:
dims: list of number of units in each layer of encoder. dims[0] is input dim, dims[-1] is units in hidden layer.
The decoder is symmetric with encoder. So number of layers of the auto-encoder is 2*len(dims)-1
act: activation, not applied to Input, Hidden and Output layers
return:
(ae_model, encoder_model), Model of autoencoder and model of encoder
"""
x = tf.keras.layers.Input(shape=(1,), dtype=tf.string)
h = tf.keras.layers.Lambda(UniversalEmbedding, output_shape=(512,))(x)
return Model(inputs=x, outputs=h, name='encoder')
class ClusteringLayer(Layer):
"""
Clustering layer converts input sample (feature) to soft label, i.e. a vector that represents the probability of the
sample belonging to each cluster. The probability is calculated with student's t-distribution.
# Example
```
model.add(ClusteringLayer(n_clusters=10))
```
# Arguments
n_clusters: number of clusters.
weights: list of Numpy array with shape `(n_clusters, n_features)` witch represents the initial cluster centers.
alpha: parameter in Student's t-distribution. Default to 1.0.
# Input shape
2D tensor with shape: `(n_samples, n_features)`.
# Output shape
2D tensor with shape: `(n_samples, n_clusters)`.
"""
def __init__(self, n_clusters, weights=None, alpha=1.0, **kwargs):
if 'input_shape' not in kwargs and 'input_dim' in kwargs:
kwargs['input_shape'] = (kwargs.pop('input_dim'),)
super(ClusteringLayer, self).__init__(**kwargs)
self.n_clusters = n_clusters
self.alpha = alpha
self.initial_weights = weights
self.input_spec = InputSpec(ndim=2)
def build(self, input_shape):
assert len(input_shape) == 2
input_dim = input_shape[1].value
self.input_spec = InputSpec(dtype=K.floatx(), shape=(None, input_dim))
self.clusters = self.add_weight(shape=(self.n_clusters, input_dim), initializer='glorot_uniform', name='clusters')
if self.initial_weights is not None:
self.set_weights(self.initial_weights)
del self.initial_weights
self.built = True
def call(self, inputs, **kwargs):
""" student t-distribution, as same as used in t-SNE algorithm.
q_ij = 1/(1+dist(x_i, u_j)^2), then normalize it.
Arguments:
inputs: the variable containing data, shape=(n_samples, n_features)
Return:
q: student's t-distribution, or soft labels for each sample. shape=(n_samples, n_clusters)
"""
q = 1.0 / (1.0 + (K.sum(K.square(K.expand_dims(inputs,axis=1) - self.clusters), axis=2) / self.alpha))
q **= (self.alpha + 1.0) / 2.0
q = K.transpose(K.transpose(q) / K.sum(q, axis=1))
return q
def compute_output_shape(self, input_shape):
assert input_shape and len(input_shape) == 2
return input_shape[0], self.n_clusters
def get_config(self):
config = {'n_clusters': self.n_clusters}
base_config = super(ClusteringLayer, self).get_config()
return dict(list(base_config.items()) + list(config.items()))
class DEC(object):
def __init__(self,
dims,
n_clusters=10,
alpha=1.0,
init='glorot_uniform'):
super(DEC, self).__init__()
self.dims = dims
self.input_dim = dims[0]
self.n_stacks = len(self.dims) - 1
self.n_clusters = n_clusters
self.alpha = alpha
self.encoder = autoencoder(self.dims, init=init)
# prepare DEC model
clustering_layer = ClusteringLayer(self.n_clusters, name='clustering')(self.encoder.output)
self.model = Model(inputs=self.encoder.input, outputs=clustering_layer)
def load_weights(self, weights): # load weights of DEC model
self.model.load_weights(weights)
def extract_features(self, x):
return self.encoder.predict(x)
def predict(self, x): # predict cluster labels using the output of clustering layer
q = self.model.predict(x, verbose=0)
return q.argmax(1)
@staticmethod
def target_distribution(q):
weight = q ** 2 / q.sum(0)
return (weight.T / weight.sum(1)).T
def compile(self, optimizer='sgd', loss='kld'):
self.model.compile(optimizer=optimizer, loss=loss)
def fit(self, x, y=None, maxiter=2e4, batch_size=256, tol=1e-3,
update_interval=140, save_dir='./results/temp'):
print('Update interval', update_interval)
save_interval = int(x.shape[0] / batch_size) * 5 # 5 epochs
print('Save interval', save_interval)
# Step 1: initialize cluster centers using k-means
t1 = time()
print('Initializing cluster centers with k-means.')
kmeans = KMeans(n_clusters=self.n_clusters, n_init=20)
y_pred = kmeans.fit_predict(self.encoder.predict(x))
y_pred_last = np.copy(y_pred)
self.model.get_layer(name='clustering').set_weights([kmeans.cluster_centers_])
# Step 2: deep clustering
# logging file
import csv
logfile = open(save_dir + '/dec_log.csv', 'w')
logwriter = csv.DictWriter(logfile, fieldnames=['iter', 'acc', 'nmi', 'ari', 'loss'])
logwriter.writeheader()
loss = 0
index = 0
index_array = np.arange(x.shape[0])
for ite in range(int(maxiter)):
if ite % update_interval == 0:
q = self.model.predict(x, verbose=0)
p = self.target_distribution(q) # update the auxiliary target distribution p
# evaluate the clustering performance
y_pred = q.argmax(1)
if y is not None:
acc = np.round(metrics.acc(y, y_pred), 5)
nmi = np.round(metrics.nmi(y, y_pred), 5)
ari = np.round(metrics.ari(y, y_pred), 5)
loss = np.round(loss, 5)
logdict = dict(iter=ite, acc=acc, nmi=nmi, ari=ari, loss=loss)
logwriter.writerow(logdict)
print('Iter %d: acc = %.5f, nmi = %.5f, ari = %.5f' % (ite, acc, nmi, ari), ' ; loss=', loss)
# check stop criterion
delta_label = np.sum(y_pred != y_pred_last).astype(np.float32) / y_pred.shape[0]
y_pred_last = np.copy(y_pred)
if ite > 0 and delta_label < tol:
print('delta_label ', delta_label, '< tol ', tol)
print('Reached tolerance threshold. Stopping training.')
logfile.close()
break
# train on batch
# if index == 0:
# np.random.shuffle(index_array)
idx = index_array[index * batch_size: min((index+1) * batch_size, x.shape[0])]
loss = self.model.train_on_batch(x=x[idx], y=p[idx])
index = index + 1 if (index + 1) * batch_size <= x.shape[0] else 0
# save intermediate model
if ite % save_interval == 0:
print('saving model to:', save_dir + '/DEC_model_' + str(ite) + '.h5')
self.model.save_weights(save_dir + '/DEC_model_' + str(ite) + '.h5')
ite += 1
# save the trained model
logfile.close()
print('saving model to:', save_dir + '/DEC_model_final.h5')
self.model.save_weights(save_dir + '/DEC_model_final.h5')
return y_pred
if __name__ == "__main__":
# load dataset
from datasets import load_retures_keras
x, y = load_retures_keras(text=True)
n_clusters = len(np.unique(y))
init = 'glorot_uniform'
update_interval = 30
batch_size = 256
maxiter = 2e5
tol=0.0001
save_dir = 'results'
# prepare the DEC model
dec = DEC(dims=[x.shape[-1], 500, 500, 2000, 10], n_clusters=n_clusters, init=init)
dec.model.summary()
t0 = time()
dec.compile('adam', loss='kld')
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
sess.run(tf.tables_initializer())
y_pred = dec.fit(x, y=y, tol=tol, maxiter=maxiter, batch_size=batch_size,
update_interval=update_interval, save_dir=save_dir)
print('acc:', metrics.acc(y, y_pred))
print('clustering time: ', (time() - t0))