class ssDA(object):
"""Semantic Stacked denoising auto-encoder class (SdA)
A stacked denoising autoencoder model is obtained by stacking several
dAs. The hidden layer of the dA at layer `i` becomes the input of
the dA at layer `i+1`. The first layer dA gets as input the input of
the SdA, and the hidden layer of the last dA represents the output.
Note that after pretraining, the SdA is dealt with as a normal MLP,
the dAs are only used to initialize the weights.
Semantic distortion comes from an additional MLP layer tacked on after reconstruction by the autoencoding chain.
"""
def dump(self, f_dump):
#f_dump: file object
# dump parameters in the form of a dictionary to this file.
#params of relevance: sigmoid and out_sigmoid's W and b.
# also, n_in and n_out for each.
hidden_paramdicts = []
for layer in (self.sigmoid_layers + self.out_sigmoid_layers):
hidden_paramdicts.append({
'W': layer.W.get_value(),
'b': layer.b.get_value(),
'n_in': layer.n_in,
'n_out': layer.n_out
})
pickle.dump(hidden_paramdicts, f_dump)
f_dump.close()
def load(self, f_load):
#f_load: file object
# laod parameters in the form of a dictionary from this file.
#params of relevance: sigmoid and out_sigmoid's W and b
# also, n_in and n_out for each.
hidden_paramdicts = pickle.load(f_load)
for (hidden_paramdict,
layer) in zip(hidden_paramdicts,
self.sigmoid_layers + self.out_sigmoid_layers):
layer.W.set_value(hidden_paramdict['W'].astype('float32'))
layer.b.set_value(hidden_paramdict['b'].astype('float32'))
layer.n_in = hidden_paramdict['n_in']
layer.n_out = hidden_paramdict['n_out']
f_load.close()
def __init__(self,
numpy_rng,
f_load_MLP=None,
f_load_SDA=None,
theano_rng=None,
n_ins=784,
hidden_layers_sizes=[500, 500],
n_outs=10,
corruption_levels=[0.1, 0.1],
name_appendage='',
xtropy_fraction=0):
""" This class is made to support a variable number of layers.
:type numpy_rng: numpy.random.RandomState
:param numpy_rng: numpy random number generator used to draw initial
weights
:type theano_rng: theano.tensor.shared_randomstreams.RandomStreams
:param theano_rng: Theano random generator; if None is given one is
generated based on a seed drawn from `rng`
:type n_ins: int
:param n_ins: dimension of the input to the sdA
:type n_layers_sizes: list of ints
:param n_layers_sizes: intermediate layers size, must contain
at least one value
:type n_outs: int
:param n_outs: dimension of the output of the network
:type corruption_levels: list of float
:param corruption_levels: amount of corruption to use for each
layer
"""
self.sigmoid_layers = []
self.out_sigmoid_layers = []
self.dA_layers = []
self.params = []
self.n_layers = len(hidden_layers_sizes)
assert self.n_layers > 0
if not theano_rng:
theano_rng = RandomStreams(numpy_rng.randint(2**30))
# allocate symbolic variables for the data
self.x = T.matrix('x') # the data is presented as rasterized images
self.y = T.ivector('y') # the labels are presented as 1D vector of
# [int] labels
for i in xrange(self.n_layers):
# the size of the input is either the number of hidden units of
# the layer below or the input size if we are on the first layer
if i == 0:
input_size = n_ins
else:
input_size = hidden_layers_sizes[i - 1]
# the input to this layer is either the activation of the hidden
# layer below or the input of the SdA if you are on the first
# layer
if i == 0:
layer_input = self.x
else:
layer_input = self.sigmoid_layers[-1].output
sigmoid_layer = HiddenLayer_ReLU(rng=numpy_rng,
input=layer_input,
n_in=input_size,
n_out=hidden_layers_sizes[i],
name_appendage=name_appendage +
'_sigmoid_' + str(i))
# add the layer to our list of layers
self.sigmoid_layers.append(sigmoid_layer)
# its arguably a philosophical question...
# but we are going to only declare that the parameters of the
# sigmoid_layers are parameters of the StackedDAA
# the visible biases in the dA are parameters of those
# dA, but not the SdA
self.params.extend(sigmoid_layer.params)
for i in xrange(self.n_layers):
all_layers = self.sigmoid_layers + self.out_sigmoid_layers
input_size = all_layers[-1].n_out
output_size = self.sigmoid_layers[-i - 1].n_in
# the input to the inverse sigmoid layer is always the activation of the
# sigmoid layer behind it (forward sigmoid if its' the first inverse layer)
layer_input = all_layers[-1].output
out_sigmoid_layer = HiddenLayer_ReLU(
rng=numpy_rng,
input=layer_input,
n_in=input_size,
n_out=output_size,
name_appendage=name_appendage + '_outsigmoid_' + str(i))
self.out_sigmoid_layers.append(out_sigmoid_layer)
self.params.extend(out_sigmoid_layer.params)
for i in xrange(self.n_layers):
sigmoid_layer = self.sigmoid_layers[i]
# Construct a denoising autoencoder that shared weights with each layer
dA_layer = dA(numpy_rng=numpy_rng,
theano_rng=theano_rng,
input=sigmoid_layer.input,
n_visible=sigmoid_layer.n_in,
n_hidden=sigmoid_layer.n_out,
W=sigmoid_layer.W,
bhid=sigmoid_layer.b,
name_appendage=name_appendage + '_dA_' + str(i))
self.dA_layers.append(dA_layer)
if f_load_MLP != None:
self.predictLayer = MLP(rng=numpy_rng,
input=self.out_sigmoid_layers[-1].output,
f_load=f_load_MLP,
name_appendage=name_appendage +
'_MLPLayer')
elif f_load_SDA != None:
self.predictLayer = SdA(numpy_rng=numpy_rng,
n_ins=28 * 28,
hidden_layers_sizes=[1000, 1000, 1000],
n_outs=10,
input=self.out_sigmoid_layers[-1].output)
self.predictLayer.load(f_load_SDA)
self.xtropy_cost = -T.mean(
self.x * T.log(self.out_sigmoid_layers[-1].output) +
(1 - self.x) * T.log(1 - self.out_sigmoid_layers[-1].output))
self.mse_cost = T.mean(
(self.x - self.out_sigmoid_layers[-1].output)**2)
self.logloss_cost = self.predictLayer.logLayer.negative_log_likelihood(
self.y)
self.finetune_cost = xtropy_fraction * self.mse_cost + (
1 - xtropy_fraction) * self.logloss_cost
self.errors = self.predictLayer.logLayer.errors(self.y)
def pretraining_functions(self, train_set_x, batch_size):
''' Generates a list of functions, each of them implementing one
step in trainnig the dA corresponding to the layer with same index.
The function will require as input the minibatch index, and to train
a dA you just need to iterate, calling the corresponding function on
all minibatch indexes.
:type train_set_x: theano.tensor.TensorType
:param train_set_x: Shared variable that contains all datapoints used
for training the dA
:type batch_size: int
:param batch_size: size of a [mini]batch
:type learning_rate: float
:param learning_rate: learning rate used during training for any of
the dA layers
'''
# index to a [mini]batch
index = T.lscalar('index') # index to a minibatch
corruption_level = T.scalar('corruption') # % of corruption to use
learning_rate = T.scalar('lr') # learning rate to use
# begining of a batch, given `index`
batch_begin = index * batch_size
# ending of a batch given `index`
batch_end = batch_begin + batch_size
pretrain_fns = []
for dA in self.dA_layers:
# get the cost and the updates list
cost, updates = dA.get_cost_updates_ReLU(corruption_level,
learning_rate)
# compile the theano function
fn = theano.function(
inputs=[
index,
theano.Param(corruption_level, default=0.2),
theano.Param(learning_rate, default=0.1)
],
outputs=cost,
updates=updates,
givens={self.x: train_set_x[batch_begin:batch_end]})
# append `fn` to the list of functions
pretrain_fns.append(fn)
return pretrain_fns
def build_finetune_functions(self, datasets, batch_size, learning_rate):
'''Generates a function `train` that implements one step of
finetuning, a function `validate` that computes the error on
a batch from the validation set, and a function `test` that
computes the error on a batch from the testing set
:type datasets: list of pairs of theano.tensor.TensorType
:param datasets: It is a list that contain all the datasets;
the has to contain three pairs, `train`,
`valid`, `test` in this order, where each pair
is formed of two Theano variables, one for the
datapoints, the other for the labels
:type batch_size: int
:param batch_size: size of a minibatch
:type learning_rate: float
:param learning_rate: learning rate used during finetune stage
'''
(train_set_x, train_set_y) = datasets[0]
(valid_set_x, valid_set_y) = datasets[1]
(test_set_x, test_set_y) = datasets[2]
# compute number of minibatches for training, validation and testing
n_valid_batches = valid_set_x.get_value(borrow=True).shape[0]
n_valid_batches /= batch_size
n_test_batches = test_set_x.get_value(borrow=True).shape[0]
n_test_batches /= batch_size
index = T.lscalar('index') # index to a [mini]batch
# compute the gradients with respect to the model parameters
gparams = T.grad(self.finetune_cost, self.params)
# compute list of fine-tuning updates
updates = [(param, param - gparam * learning_rate)
for param, gparam in zip(self.params, gparams)]
train_fn = theano.function(
inputs=[index],
outputs=self.finetune_cost,
updates=updates,
givens={
self.x:
train_set_x[index * batch_size:(index + 1) * batch_size],
self.y:
train_set_y[index * batch_size:(index + 1) * batch_size]
},
name='train')
test_score_i = theano.function(
[index],
self.errors,
givens={
self.x:
test_set_x[index * batch_size:(index + 1) * batch_size],
self.y: test_set_y[index * batch_size:(index + 1) * batch_size]
},
name='test')
valid_score_i = theano.function(
[index],
self.errors,
givens={
self.x:
valid_set_x[index * batch_size:(index + 1) * batch_size],
self.y:
valid_set_y[index * batch_size:(index + 1) * batch_size]
},
name='valid')
valid_xtropy_logloss_i = theano.function(
inputs=[index],
outputs=[self.xtropy_cost, self.logloss_cost],
givens={
self.x:
valid_set_x[index * batch_size:(index + 1) * batch_size],
self.y:
valid_set_y[index * batch_size:(index + 1) * batch_size]
},
name='valid_xtropy_logloss')
# Create a function that scans the entire validation set
def valid_score():
return [valid_score_i(i) for i in xrange(n_valid_batches)]
def valid_xtropy_logloss():
return [valid_xtropy_logloss_i(i) for i in xrange(n_valid_batches)]
# Create a function that scans the entire test set
def test_score():
return [test_score_i(i) for i in xrange(n_test_batches)]
return train_fn, valid_score, test_score, valid_xtropy_logloss
class ssDA(object):
"""Semantic Stacked denoising auto-encoder class (SdA)
A stacked denoising autoencoder model is obtained by stacking several
dAs. The hidden layer of the dA at layer `i` becomes the input of
the dA at layer `i+1`. The first layer dA gets as input the input of
the SdA, and the hidden layer of the last dA represents the output.
Note that after pretraining, the SdA is dealt with as a normal MLP,
the dAs are only used to initialize the weights.
Semantic distortion comes from an additional MLP layer tacked on after reconstruction by the autoencoding chain.
"""
def dump(self,f_dump):
#f_dump: file object
# dump parameters in the form of a dictionary to this file.
#params of relevance: sigmoid and out_sigmoid's W and b.
# also, n_in and n_out for each.
hidden_paramdicts = []
for layer in (self.sigmoid_layers + self.out_sigmoid_layers):
hidden_paramdicts.append({'W': layer.W.get_value(),
'b': layer.b.get_value(),
'n_in': layer.n_in,
'n_out': layer.n_out})
pickle.dump(hidden_paramdicts,f_dump)
f_dump.close()
def load(self,f_load):
#f_load: file object
# laod parameters in the form of a dictionary from this file.
#params of relevance: sigmoid and out_sigmoid's W and b
# also, n_in and n_out for each.
hidden_paramdicts = pickle.load(f_load)
for (hidden_paramdict,layer) in zip(hidden_paramdicts,self.sigmoid_layers+self.out_sigmoid_layers):
layer.W.set_value(hidden_paramdict['W'].astype('float32'))
layer.b.set_value(hidden_paramdict['b'].astype('float32'))
layer.n_in = hidden_paramdict['n_in']
layer.n_out = hidden_paramdict['n_out']
f_load.close()
def __init__(
self,
numpy_rng,
f_load_MLP=None,
f_load_SDA=None,
theano_rng=None,
n_ins=784,
hidden_layers_sizes=[1000, 1000, 1000, 15],
n_outs=10,
corruption_levels=[0.1, 0.1],
name_appendage='',
xtropy_fraction = 1,
dropout_probs=[0.0,0.5,0.5,0.5,0.1]
):
""" This class is made to support a variable number of layers.
:type numpy_rng: numpy.random.RandomState
:param numpy_rng: numpy random number generator used to draw initial
weights
:type theano_rng: theano.tensor.shared_randomstreams.RandomStreams
:param theano_rng: Theano random generator; if None is given one is
generated based on a seed drawn from `rng`
:type n_ins: int
:param n_ins: dimension of the input to the sdA
:type n_layers_sizes: list of ints
:param n_layers_sizes: intermediate layers size, must contain
at least one value
:type n_outs: int
:param n_outs: dimension of the output of the network
:type corruption_levels: list of float
:param corruption_levels: amount of corruption to use for each
layer
"""
self.sigmoid_layers = []
self.out_sigmoid_layers = []
self.dA_layers = []
self.params = []
self.n_layers = len(hidden_layers_sizes)
assert self.n_layers > 0
if not theano_rng:
theano_rng = RandomStreams(numpy_rng.randint(2 ** 30))
# allocate symbolic variables for the data
self.x = T.matrix('x') # the data is presented as rasterized images
self.y = T.ivector('y') # the labels are presented as 1D vector of
# [int] labels
for i in xrange(self.n_layers):
# the size of the input is either the number of hidden units of
# the layer below or the input size if we are on the first layer
if i == 0:
input_size = n_ins
else:
input_size = hidden_layers_sizes[i - 1]
# the input to this layer is either the activation of the hidden
# layer below or the input of the SdA if you are on the first
# layer
if i == 0:
layer_input = self.x
else:
layer_input = self.sigmoid_layers[-1].output
sigmoid_layer = HiddenLayer_ReLU_dropout(rng=numpy_rng,
input=layer_input,
n_in=input_size,
n_out=hidden_layers_sizes[i],
dropout_rate=dropout_probs[i+1],
name_appendage = name_appendage+'_sigmoid_'+str(i))
# add the layer to our list of layers
self.sigmoid_layers.append(sigmoid_layer)
# its arguably a philosophical question...
# but we are going to only declare that the parameters of the
# sigmoid_layers are parameters of the StackedDAA
# the visible biases in the dA are parameters of those
# dA, but not the SdA
self.params.extend(sigmoid_layer.params)
for i in xrange(self.n_layers):
all_layers = self.sigmoid_layers+self.out_sigmoid_layers
input_size = all_layers[-1].n_out
output_size = self.sigmoid_layers[-i-1].n_in
# the input to the inverse sigmoid layer is always the activation of the
# sigmoid layer behind it (forward sigmoid if its' the first inverse layer)
layer_input = all_layers[-1].output
out_sigmoid_layer = HiddenLayer_ReLU_dropout(rng=numpy_rng,
input=layer_input,
n_in=input_size,
n_out=output_size,
dropout_rate=dropout_probs[-i-2],
name_appendage = name_appendage+'_outsigmoid_'+str(i))
self.out_sigmoid_layers.append(out_sigmoid_layer)
self.params.extend(out_sigmoid_layer.params)
for i in xrange(self.n_layers):
sigmoid_layer = self.sigmoid_layers[i]
# Construct a denoising autoencoder that shared weights with each layer
dA_layer = dA(numpy_rng=numpy_rng,
theano_rng=theano_rng,
input=sigmoid_layer.input,
n_visible=sigmoid_layer.n_in,
n_hidden=sigmoid_layer.n_out,
W = sigmoid_layer.W,
bhid=sigmoid_layer.b,
name_appendage=name_appendage+'_dA_'+str(i)
)
self.dA_layers.append(dA_layer)
if f_load_MLP != None:
self.predictLayer = MLP(
rng = numpy_rng,
input=self.out_sigmoid_layers[-1].output,
f_load = f_load_MLP,
name_appendage = name_appendage+'_MLPLayer'
)
elif f_load_SDA != None:
self.predictLayer = SdA(
numpy_rng = numpy_rng,
n_ins=28 * 28,
hidden_layers_sizes=[1000, 1000, 1000],
n_outs=10,
input = self.out_sigmoid_layers[-1].output
)
self.predictLayer.load(f_load_SDA)
self.xtropy_cost = -T.mean(self.x*T.log(self.out_sigmoid_layers[-1].output) + (1-self.x)*T.log(1-self.out_sigmoid_layers[-1].output))
self.mse_cost = T.mean((self.x-self.out_sigmoid_layers[-1].output)**2)
self.logloss_cost = self.predictLayer.logLayer.negative_log_likelihood(self.y)
self.finetune_cost = xtropy_fraction*self.mse_cost + (1-xtropy_fraction)*self.logloss_cost
self.errors = self.predictLayer.logLayer.errors(self.y)
def pretraining_functions(self, train_set_x, batch_size):
''' Generates a list of functions, each of them implementing one
step in trainnig the dA corresponding to the layer with same index.
The function will require as input the minibatch index, and to train
a dA you just need to iterate, calling the corresponding function on
all minibatch indexes.
:type train_set_x: theano.tensor.TensorType
:param train_set_x: Shared variable that contains all datapoints used
for training the dA
:type batch_size: int
:param batch_size: size of a [mini]batch
:type learning_rate: float
:param learning_rate: learning rate used during training for any of
the dA layers
'''
# index to a [mini]batch
index = T.lscalar('index') # index to a minibatch
corruption_level = T.scalar('corruption') # % of corruption to use
learning_rate = T.scalar('lr') # learning rate to use
# begining of a batch, given `index`
batch_begin = index * batch_size
# ending of a batch given `index`
batch_end = batch_begin + batch_size
pretrain_fns = []
for dA in self.dA_layers:
# get the cost and the updates list
cost, updates = dA.get_cost_updates_ReLU(corruption_level,
learning_rate)
# compile the theano function
fn = theano.function(
inputs=[
index,
theano.Param(corruption_level, default=0.2),
theano.Param(learning_rate, default=0.1)
],
outputs=cost,
updates=updates,
givens={
self.x: train_set_x[batch_begin: batch_end]
}
)
# append `fn` to the list of functions
pretrain_fns.append(fn)
return pretrain_fns
def build_finetune_functions(self, datasets, batch_size, learning_rate):
'''Generates a function `train` that implements one step of
finetuning, a function `validate` that computes the error on
a batch from the validation set, and a function `test` that
computes the error on a batch from the testing set
:type datasets: list of pairs of theano.tensor.TensorType
:param datasets: It is a list that contain all the datasets;
the has to contain three pairs, `train`,
`valid`, `test` in this order, where each pair
is formed of two Theano variables, one for the
datapoints, the other for the labels
:type batch_size: int
:param batch_size: size of a minibatch
:type learning_rate: float
:param learning_rate: learning rate used during finetune stage
'''
(train_set_x, train_set_y) = datasets[0]
(valid_set_x, valid_set_y) = datasets[1]
(test_set_x, test_set_y) = datasets[2]
# compute number of minibatches for training, validation and testing
n_valid_batches = valid_set_x.get_value(borrow=True).shape[0]
n_valid_batches /= batch_size
n_test_batches = test_set_x.get_value(borrow=True).shape[0]
n_test_batches /= batch_size
index = T.lscalar('index') # index to a [mini]batch
# compute the gradients with respect to the model parameters
gparams = T.grad(self.finetune_cost, self.params)
# compute list of fine-tuning updates
updates = [
(param, param - gparam * learning_rate)
for param, gparam in zip(self.params, gparams)
]
train_fn = theano.function(
inputs=[index],
outputs=self.finetune_cost,
updates=updates,
givens={
self.x: train_set_x[
index * batch_size: (index + 1) * batch_size
],
self.y: train_set_y[
index * batch_size: (index + 1) * batch_size
]
},
name='train'
)
test_score_i = theano.function(
[index],
self.errors,
givens={
self.x: test_set_x[
index * batch_size: (index + 1) * batch_size
],
self.y: test_set_y[
index * batch_size: (index + 1) * batch_size
]
},
name='test'
)
valid_score_i = theano.function(
[index],
self.errors,
givens={
self.x: valid_set_x[
index * batch_size: (index + 1) * batch_size
],
self.y: valid_set_y[
index * batch_size: (index + 1) * batch_size
]
},
name='valid'
)
valid_xtropy_logloss_i = theano.function(
inputs=[index],
outputs=[self.xtropy_cost, self.logloss_cost],
givens={
self.x: valid_set_x[
index * batch_size: (index + 1) * batch_size
],
self.y: valid_set_y[
index * batch_size: (index + 1) * batch_size
]
},
name='valid_xtropy_logloss'
)
# Create a function that scans the entire validation set
def valid_score():
return [valid_score_i(i) for i in xrange(n_valid_batches)]
def valid_xtropy_logloss():
return [valid_xtropy_logloss_i(i) for i in xrange(n_valid_batches)]
# Create a function that scans the entire test set
def test_score():
return [test_score_i(i) for i in xrange(n_test_batches)]
return train_fn, valid_score, test_score, valid_xtropy_logloss