File size: 20,901 Bytes
eb90a8a | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 | """
Module for making Keras functional Neural Network models
for similarity analysis of source code.
The module provides factory class for making
Neural Network models
"""
import sys
import tensorflow
from tensorflow.keras import models
from tensorflow.keras import layers
from ModelUtils import SeqBlockFactory, BaseModelFactory
class FuncModelFactory(object):
"""
Factory of DNNs for code classification and similarity analysis
Provides functions for creating DNNs for similarity
analysis of source code represented with sequence of tokens
If the number of lables > 1 the constructed DNN uses;
- softmax for classification
- categorical_crossentropy for loss function
othervise constructed DNN uses:
- sigmoid for classification
- binary_crossentropy for loss function
"""
def __init__(self, n_labels, regularizer = None):
"""
Initialize factory for DNNs
Parameters:
- n_labels -- number of labels (classes)
- regularizer -- regularizer as a pair (l1, l2)
or None if no regularization is applied
Same regularization is applied to all layers
"""
self.n_labels = n_labels
self.class_layer = \
BaseModelFactory.compClassification(n_labels)
self._regularizer = \
BaseModelFactory.computeRegularizer(regularizer)
def _convLayer(self, input_tensor, convolution,
activation = None, bias = True):
"""
Construct a convolutional layer
Regularization is applied with self._regularizer
Parameters:
- input_tensor -- input tensor
- convolution -- specification of the convolutional layer
as list of tuples:
either (n filters, kernel width, stride)
or (n filters, kernel width) if stride = 1
- activation -- activation on convolutional layer
- bias -- use or do not use bias on convolutional layer
Returns:
- output tensor if the layer is successfully created
- None if the layer is specified incorrectly
"""
if len(convolution) == 3:
return layers.Conv1D(convolution[0], convolution[1],
strides=convolution[2],
activation = activation,
use_bias = bias,
kernel_regularizer=self._regularizer,
kernel_initializer = BaseModelFactory.computeInitializer()
)(input_tensor)
elif len(convolution) == 2:
return layers.Conv1D(convolution[0], convolution[1],
activation = activation,
use_bias = bias,
kernel_regularizer=self._regularizer,
kernel_initializer = BaseModelFactory.computeInitializer()
)(input_tensor)
else:
return None
def _convolution1(self, input_tensor, convolutions,
activation = None, bias = True,
out_no_act = True):
"""
Make a stack of convolutional layers
Parameters:
- input_tensor -- input tensor
- convolutions -- specification of convolutional layers
as list of tuples:
either (n filters, kernel width, stride)
or (n filters, kernel width) if stride = 1
- activation -- activation on all convolutional layers
but the last one
- bias -- use or do not use bias on convolutional layers
- out_no_act -- flag of no activation on last layer
Returns:
- output tensor
"""
_output = input_tensor
for _i in range(len(convolutions)):
_act = None if _i == len(convolutions) - 1 and out_no_act \
else activation
_output = self._convLayer(_output, convolutions[_i],
activation = _act, bias = bias)
if _output is None:
print(f"Wrong specification of convolutional layer {-i}")
_n_conv_parms = len(convolutions[_i])
print(f"It has incorrect number {_n_conv_parm} of parameters")
sys.exit(0)
return _output
def _absDifference(self, input1, input2):
"""
Compute absolute value of difference of two tensor
Parameters:
- input1 -- 1-st operand
- input2 -- 2-nd operand
"""
_diff1 = layers.Subtract()([input1, input2])
_diff2 = layers.Subtract()([input2, input1])
#return layers.Maximum()([_diff1, _diff2])
return layers.Minimum()([_diff1, _diff2])
def _denseClassifier(self, input_tensor, dense, bias = True):
"""
Make a stack of dense layers with classifier
Classifier is either softmax for more than 2 classes
or sigmoid for 2 classes
It is expected that input shape is known
Parameters:
- input_tensor -- input tensor
- dense -- specification of dense layers as list of
their width except the last one
- bias -- use or do not use bias on dense layers
Returns: output tensor
"""
_output = input_tensor
if dense:
for _w in dense:
_output = layers.Dense(_w, activation = 'relu',
use_bias = bias,
kernel_regularizer=self._regularizer,
kernel_initializer = BaseModelFactory.computeInitializer()
)(_output)
_output = layers.Dense(self.n_labels,
activation = self.class_layer,
use_bias = bias,
kernel_regularizer=self._regularizer,
kernel_initializer = BaseModelFactory.computeInitializer())(_output)
return _output
def _dotProdClassifier(self, input1, input2):
"""
Add dot product classifier
Dot product is not normalized but is transformed by sigmoid
to probabilities as it helps convergence which is still inferior
to conventional methods of merging two embeddings
!!!The function is for experiments only
Parameters:
- input1 -- 1-st operand of dot product
- input2 -- 2-nd operand of dot product
"""
_output = layers.Dot(axes = 1, normalize = False)([input1, input2])
return tensorflow.keras.layers.Activation("sigmoid")(_output)
def twoWaySimilarityCNN(self, n_tokens,
convolutions, w_dense,
pool = 'max', side_dense = [],
conv_act = None, conv_bias = True,
dense_bias = True, shared = False,
input_type = "one_hot",
embedding_dim = None,
regul_dense_only = True,
dropout_rate = 0,
optimizer = "rmsprop"):
"""
Make symmetric two way convolutional dnn
for sequence similarity analysis
Parameters:
- n_tokens -- number of types of tokens
- convolutions -- specification of convolutional layers
as list of tuples:
either (kernel width, n filters, stride)
or (kernel width, n filters) if stride = 1
- w_dense -- specification of dense layers as list of
their width except the last one
- pool -- pooling operation either:
'max' for maximum or
'aver' for average
- side_dense -- specification of each branch dense layers
as list of their width
- conv_act -- activation of convolutional layers
- conv_bias -- use bias for convolutional layers
- dense_bias -- use bias for dense layers
- input_type -- type of input layer:
- one_hot - one hot coding of input samples
- categorical - categorical coding
decoded by preprocessing layer
- trainable - categorical coding processed
by trainable embedding layer
- embedding_dim -- size of embedding vectors
- dropout_rate -- dropout rate for inserted dropout layer
If it is 0 or None no droput layer is inserted
- regul_dense_only -- flag to regulirize only dense layers
- optimizer -- training optimizer
Returns compiled dnn
"""
if input_type == "one_hot":
_input1 = layers.Input(shape=(None, n_tokens,))
_input2 = layers.Input(shape=(None, n_tokens,))
elif input_type == "categorical" or input_type == "trainable":
_input1 = layers.Input(shape=(None,))
_input2 = layers.Input(shape=(None,))
else:
sys.exit(f"Wrong type {input_type} of input_type parameter of twoWaySimilarityCNN")
_maker = SeqBlockFactory(n_tokens)
if shared:
_conv = _maker.convBlock(
convolutions, dense = side_dense, pool = pool,
conv_act = conv_act, conv_bias = conv_bias,
dense_bias = dense_bias, input_type = input_type,
regularizer = self._regularizer,
embedding_dim = embedding_dim,
dropout_rate = dropout_rate,
regul_dense_only = regul_dense_only)
_pool1 = _conv(_input1)
_pool2 = _conv(_input2)
else:
_conv1 = _maker.convBlock(
convolutions, dense = side_dense, pool = pool,
conv_act = conv_act, conv_bias = conv_bias,
dense_bias = dense_bias, input_type = input_type,
regularizer = self._regularizer,
embedding_dim = embedding_dim,
dropout_rate = dropout_rate,
regul_dense_only = regul_dense_only)
_conv2 = _maker.convBlock(
convolutions, dense = side_dense, pool = pool,
conv_act = conv_act, conv_bias = conv_bias,
dense_bias = dense_bias, input_type = input_type,
regularizer = self._regularizer,
embedding_dim = embedding_dim,
dropout_rate = dropout_rate,
regul_dense_only = regul_dense_only)
_pool1 = _conv1(_input1)
_pool2 = _conv2(_input2)
_merged = layers.concatenate([_pool1, _pool2], axis=-1)
_output = self._denseClassifier(_merged, w_dense)
return self._compile([_input1, _input2,], _output,
optimizer = optimizer)
def _compile(self, inputs, output, optimizer = "rmsprop"):
"""
Make functional dnn model and compile it
Parameters:
- inputs -- list of dnn inputs
- output -- dnn output
- optimizer -- training optimizer
Returns compiled DNN
"""
_dnn = models.Model(inputs, output)
_loss = BaseModelFactory.compLoss(self.n_labels)
_dnn = SeqBlockFactory.compileDNN(
_dnn, _loss, optimizer = optimizer)
return _dnn
def siameseSimilarityCNN(self, n_tokens,
convolutions, w_dense,
pool = 'max', side_dense = [],
conv_act = None, conv_bias = True,
dense_bias = True,
input_type = "trainable",
embedding_dim = None,
regul_dense_only = True,
dropout_rate = 0,
merge = "concatenate",
optimizer = "adam"):
"""
Make two way convolutional Siamese dnn
for sequence similarity analysis
Parameters:
- n_tokens -- number of types of tokens
- convolutions -- specification of convolutional layers
as list of tuples:
either (kernel width, n filters, stride)
or (kernel width, n filters) if stride = 1
- w_dense -- specification of dense layers as list of
their width except the last one
- pool -- pooling operation either:
'max' for maximum or
'aver' for average
- side_dense -- specification of each branch dense layers
as list of their width
- conv_act -- activation of convolutional layers
- conv_bias -- use bias for convolutional layers
- dense_bias -- use bias for dense layers
- input_type -- type of input layer:
- categorical - categorical coding
decoded by preprocessing layer
- trainable - categorical coding processed
by trainable embedding layer
- embedding_dim -- size of embedding vectors
- dropout_rate -- dropout rate for inserted dropout layer
If it is 0 or None no droput layer is inserted
- merge -- method of merging outputs of Siamese blocks
* if concatenate, the outputs are concatenated
and processed with dense layers
* if subract, absolute value of difference
of the outputs is computed
* if "dot_prod_sigmoid" it is computed
sigmoid of dot product of outputs
* if "cosine" it is computed
normalized dot product of normalized outputs
- regul_dense_only -- flag to regulirize only dense layers
- optimizer -- training optimizer
Returns compiled dnn
"""
_input1 = layers.Input(shape=(None,))
_input2 = layers.Input(shape=(None,))
_maker = SeqBlockFactory(n_tokens)
_conv = _maker.convBlock(
convolutions, dense = side_dense, pool = pool,
conv_act = conv_act, conv_bias = conv_bias,
dense_bias = dense_bias, input_type = input_type,
regularizer = self._regularizer,
embedding_dim = embedding_dim,
dropout_rate = dropout_rate,
regul_dense_only = regul_dense_only)
_pool1 = _conv(_input1)
_pool2 = _conv(_input2)
if merge == "concatenate":
_merged = layers.concatenate([_pool1, _pool2])
_output = self._denseClassifier(_merged, w_dense)
elif merge == "subtract":
_merged = self._absDifference(_pool1, _pool2)
_output = self._denseClassifier(_merged, w_dense)
elif merge == "dot_prod_sigmoid":
_output = self._dotProdClassifier(_pool1, _pool2)
elif merge == "cosine":
_output = layers.Dot(axes = 1,
normalize = True)([_pool1, _pool2])
else:
sys.exit(f"Wrong method {merge} of merging outputs of Siamese blocks")
return self._compile([_input1, _input2,], _output,
optimizer = optimizer)
def twoWayAsymSimilarityCNN(self, n_tokens1, n_tokens2,
convolutions, w_dense,
pool = 'max', side_dense = [],
conv_act = None, conv_bias = True,
dense_bias = True,
input_type = "trainable",
embedding_dim1 = None,
embedding_dim2 = None,
regul_dense_only = True,
dropout_rate = 0,
merge = "subtract",
optimizer = "adam"):
"""
Make asymmetric two way convolutional dnn
for similarity analysis of token sequences
of code written in two different languages
Parameters:
- n_tokens1 -- number of types of tokens for 1-st language
- n_tokens2 -- number of types of tokens for 2-nd language
- convolutions -- specification of convolutional layers
as list of tuples:
either (kernel width, n filters, stride)
or (kernel width, n filters) if stride = 1
- w_dense -- specification of dense layers as list of
their width except the last one
- pool -- pooling operation either:
'max' for maximum or
'aver' for average
- side_dense -- specification of each branch dense layers
as list of their width
- conv_act -- activation of convolutional layers
- conv_bias -- use bias for convolutional layers
- dense_bias -- use bias for dense layers
- input_type -- type of input layer:
- categorical - categorical coding
decoded by preprocessing layer
- trainable - categorical coding processed
by trainable embedding layer
- embedding_dim1 -- size of embedding vectors for 1-st language
- embedding_dim2 -- size of embedding vectors for 2-nd language
- dropout_rate -- dropout rate for inserted dropout layer
If it is 0 or None no droput layer is inserted
- merge -- method of merging outputs of Siamese blocks
* if concatenate, the outputs are concatenated
and processed with dense layers
* if subract, absolute value of difference
of the outputs is computed
* if "dot_prod_sigmoid" it is computed
sigmoid of dot product of outputs
* if "cosine" it is computed
normalized dot product of normalized outputs
- regul_dense_only -- flag to regulirize only dense layers
- optimizer -- training optimizer
Returns compiled dnn
"""
_input1 = layers.Input(shape=(None,))
_input2 = layers.Input(shape=(None,))
_maker1 = SeqBlockFactory(n_tokens1)
_conv1 = _maker1.convBlock(
convolutions, dense = side_dense, pool = pool,
conv_act = conv_act, conv_bias = conv_bias,
dense_bias = dense_bias, input_type = input_type,
regularizer = self._regularizer,
embedding_dim = embedding_dim1,
dropout_rate = dropout_rate,
regul_dense_only = regul_dense_only)
_maker2 = SeqBlockFactory(n_tokens2)
_conv2 = _maker2.convBlock(
convolutions, dense = side_dense, pool = pool,
conv_act = conv_act, conv_bias = conv_bias,
dense_bias = dense_bias, input_type = input_type,
regularizer = self._regularizer,
embedding_dim = embedding_dim2,
dropout_rate = dropout_rate,
regul_dense_only = regul_dense_only)
_pool1 = _conv1(_input1)
_pool2 = _conv2(_input2)
if merge == "concatenate":
_merged = layers.concatenate([_pool1, _pool2], axis=-1)
_output = self._denseClassifier(_merged, w_dense)
elif merge == "subtract":
_merged = self._absDifference(_pool1, _pool2)
_output = self._denseClassifier(_merged, w_dense)
elif merge == "dot_prod_sigmoid":
_output = self._dotProdClassifier(_pool1, _pool2)
elif merge == "cosine":
_output = layers.Dot(axes = 1,
normalize = True)([_pool1, _pool2])
else:
sys.exit(f"Wrong method {merge} of merging outputs of Siamese blocks")
return self._compile([_input1, _input2,], _output,
optimizer = optimizer)
#---------------- End of class FuncModelFactory ---------------------------
|