def evaluate_lenet5(learning_rate=0.1, n_epochs=200,
dataset='mnist.pkl.gz',
nkerns=[20,50],batch_size=500):
"""Demonstates lenet on MNIST dataset
"""
rng = numpy.random.RandomState(1234)
print('Loading Data'+'.'*20)
datasets = load_data(dataset)
trainSetX, trainSetY = datasets[0]
validSetX, validSetY = datasets[1]
testSetX, testSetY = datasets[2]
n_train_batches = trainSetX.get_value(borrow=True).shape[0] // batch_size
n_valid_batches = validSetX.get_value(borrow=True).shape[0] // batch_size
n_test_batches = testSetX.get_value(borrow=True).shape[0] // batch_size
print('Building Data'+'.'*20)
index = T.lscalar('index')
x = T.matrix('x')
y = T.ivector('y')
# Reshape matrix of rasterized images of shape (batch_size, 28*28)
# to a 4D tensor, compatible with our LeNetConvPoolLayer
layer0_input = x.reshape((batch_size,1,28,28))
# construct the first convolutional pooling layer
# filtering reduces the image size to (28-5+1,28-5+1) = (24, 24)
# maxpooling reduces this further to (24/2, 24/2) = (12, 12)
# 4D output tensor is thus of shape (batch_size,nkerns[0],12,12)
layer0 = LeNetConvPoolLayer(
rng=rng,
input=layer0_input,
image_shape=(batch_size,1,28,28),
filter_shape=(nkerns[0],1,5,5),
poolsize=(2,2)
)
# construct the second convolutional pooling layer
# filtering reduces the image size to (12-5+1,12-5+1) = (8, 8)
# maxpooling reduces this further to (8/2, 8/2) = (4, 4)
# 4D output tensor is thus of shape (batch_size,nkerns[1],4,4)
layer1 = LeNetConvPoolLayer(
rng=rng,
input=layer0.output,
image_shape=(batch_size,nkerns[0],12,12),
filter_shape=(nkerns[1],nkerns[0],5,5),
poolsize=(2,2)
)
layer2_input = layer1.output.flatten(2)
layer2 = HiddenLayer(
rng=rng,
input=layer2_input,
n_in=nkerns[1]*4*4,
n_out=500,
activation=T.tanh
)
layer3 = LogisticRegression(input=layer2.output,n_in=500,n_out=10)
testModel = theano.function(
inputs=[index],
outputs=layer3.errors(y),
givens={
x:testSetX[index*batch_size:(index+1)*batch_size],
y:testSetY[index*batch_size:(index+1)*batch_size]
}
)
validModel = theano.function(
inputs=[index],
outputs=layer3.errors(y),
givens={
x:validSetX[index*batch_size:(index+1)*batch_size],
y:validSetY[index*batch_size:(index+1)*batch_size]
}
)
params = layer3.params+layer2.params+layer1.params+layer0.params
cost = layer3.negative_log_likelihood(y)
grads = T.grad(cost,params)
updates= [(param, param - learning_rate*grad)
for param,grad in zip(params,grads)
]
trainModel = theano.function(
inputs=[index],
outputs=cost,
updates=updates,
givens={
x:trainSetX[index*batch_size:(index+1)*batch_size],
y:trainSetY[index*batch_size:(index+1)*batch_size]
}
)
print('Training'+'.'*20)
patience = 10000
patience_increase = 2
improvement_threshold = 2
validation_frequence = min(n_train_batches, patience/2)
best_validation_loss = numpy.inf
best_iter = 0
test_score =0.
start_time = timeit.default_timer()
epoch = 0
done_looping = False
while (epoch < n_epochs) and (not done_looping):
epoch += 1
for mini_batch_index in range(n_train_batches):
iter = (epoch - 1) * n_train_batches + mini_batch_index
if iter % 100 == 0:
print('training @ iter = ' , iter)
cost_ij = trainModel(mini_batch_index)
if (iter + 1) % validation_frequence ==0:
validation_losses = [validModel(i) for i
in range(n_valid_batches)]
this_validation_losses = numpy.mean(validation_losses)
print('epoch %i, minibatch %i/%i, validation error %f %%' %
(epoch, mini_batch_index+1, n_train_batches,
this_validation_losses*100)
)
if this_validation_losses < best_validation_loss:
best_validation_loss = this_validation_losses
best_iter = iter
if this_validation_losses < best_validation_loss * \
improvement_threshold:
patience = max(patience, patience*patience_increase)
test_losses = [testModel(i) for i in range(n_test_batches)]
test_score = numpy.mean(test_losses)
print(' epoch %i, minibatch %i/%i, test error of'
'best model %f %%'%
(epoch, mini_batch_index+1, n_train_batches,
this_validation_losses*100)
)
if patience <= iter:
done_looping = True
break
endtime = timeit.default_timer()
print('Optimization complete.')
print('Best validation score of %f %% obtained at iteration %i, '
' with test performance %f %%' %
(best_validation_loss*100, best_iter+1, test_score*100.)
)
print('The code for file ' + os.path.split(__file__)[1]+
' ran for %.2fm' % (endtime - start_time)/60.
)
def test_rbm(learning_rate=0.1, training_epochs=15,
dataset='mnist.pkl.gz', batch_size=20,
n_chains=20, n_samples=10, output_folder='rbm_plots',
n_hidden=500):
"""
Demonstrate how to train and afterwards sample from it using Theano.
This is demonstrated on MNIST.
:param learning_rate: learning rate used for training the RBM
:param training_epochs: number of epochs used for training
:param dataset: path the the pickled dataset
:param batch_size: size of a batch used to train the RBM
:param n_chains: number of parallel Gibbs chains to be used for sampling
:param n_samples: number of samples to plot for each chain
"""
datasets = load_data(dataset)
train_set_x, train_set_y = datasets[0]
test_set_x, test_set_y = datasets[2]
# compute number of minibatches for training, validation and testing
n_train_batches = train_set_x.get_value(borrow=True).shape[0] / batch_size
# allocate symbolic variables for the data
index = T.lscalar() # index to a [mini]batch
x = T.matrix('x') # the data is presented as rasterized images
rng = numpy.random.RandomState(123)
theano_rng = RandomStreams(rng.randint(2 ** 30))
# initialize storage for the persistent chain (state = hidden
# layer of chain)
persistent_chain = theano.shared(numpy.zeros((batch_size, n_hidden),
dtype=theano.config.floatX),
borrow=True)
# construct the RBM class
rbm = RBM(input=x, n_visible=28 * 28,
n_hidden=n_hidden, numpy_rng=rng, theano_rng=theano_rng)
# get the cost and the gradient corresponding to one step of CD-15
cost, updates = rbm.get_cost_updates(lr=learning_rate,
persistent=persistent_chain, k=15)
#################################
# Training the RBM #
#################################
if not os.path.isdir(output_folder):
os.makedirs(output_folder)
os.chdir(output_folder)
# start-snippet-5
# it is ok for a theano function to have no output
# the purpose of train_rbm is solely to update the RBM parameters
train_rbm = theano.function(
[index],
cost,
updates=updates,
givens={
x: train_set_x[index * batch_size: (index + 1) * batch_size]
},
name='train_rbm'
)
plotting_time = 0.
start_time = timeit.default_timer()
# go through training epochs
for epoch in xrange(training_epochs):
# go through the training set
mean_cost = []
for batch_index in xrange(n_train_batches):
mean_cost += [train_rbm(batch_index)]
print 'Training epoch %d, cost is ' % epoch, numpy.mean(mean_cost)
# Plot filters after each training epoch
plotting_start = timeit.default_timer()
# Construct image from the weight matrix
image = Image.fromarray(
tile_raster_images(
X=rbm.W.get_value(borrow=True).T,
img_shape=(28, 28),
tile_shape=(10, 10),
tile_spacing=(1, 1)
)
)
image.save('filters_at_epoch_%i.png' % epoch)
plotting_stop = timeit.default_timer()
plotting_time += (plotting_stop - plotting_start)
end_time = timeit.default_timer()
pretraining_time = (end_time - start_time) - plotting_time
print ('Training took %f minutes' % (pretraining_time / 60.))
# end-snippet-5 start-snippet-6
#################################
# Sampling from the RBM #
#################################
# find out the number of test samples
number_of_test_samples = test_set_x.get_value(borrow=True).shape[0]
# pick random test examples, with which to initialize the persistent chain
test_idx = rng.randint(number_of_test_samples - n_chains)
persistent_vis_chain = theano.shared(
numpy.asarray(
test_set_x.get_value(borrow=True)[test_idx:test_idx + n_chains],
dtype=theano.config.floatX
)
)
# end-snippet-6 start-snippet-7
plot_every = 1000
# define one step of Gibbs sampling (mf = mean-field) define a
# function that does `plot_every` steps before returning the
# sample for plotting
(
[
presig_hids,
hid_mfs,
hid_samples,
presig_vis,
vis_mfs,
vis_samples
],
updates
) = theano.scan(
rbm.gibbs_vhv,
outputs_info=[None, None, None, None, None, persistent_vis_chain],
n_steps=plot_every
)
# add to updates the shared variable that takes care of our persistent
# chain :.
updates.update({persistent_vis_chain: vis_samples[-1]})
# construct the function that implements our persistent chain.
# we generate the "mean field" activations for plotting and the actual
# samples for reinitializing the state of our persistent chain
sample_fn = theano.function(
[],
[
vis_mfs[-1],
vis_samples[-1]
],
updates=updates,
name='sample_fn'
)
# create a space to store the image for plotting ( we need to leave
# room for the tile_spacing as well)
image_data = numpy.zeros(
(29 * n_samples + 1, 29 * n_chains - 1),
dtype='uint8'
)
for idx in xrange(n_samples):
# generate `plot_every` intermediate samples that we discard,
# because successive samples in the chain are too correlated
vis_mf, vis_sample = sample_fn()
print ' ... plotting sample ', idx
image_data[29 * idx:29 * idx + 28, :] = tile_raster_images(
X=vis_mf,
img_shape=(28, 28),
tile_shape=(1, n_chains),
tile_spacing=(1, 1)
)
# construct image
image = Image.fromarray(image_data)
image.save('samples.png')
# end-snippet-7
os.chdir('../')
def test_DBN(finetune_lr=0.1, pretraining_epochs=100,
pretrain_lr=0.01, k=1, training_epochs=1000,
dataset='d:/mnist.pkl.gz', batch_size=500):
"""
Demonstrates how to train and test a Deep Belief Network.
This is demonstrated on MNIST.
:type learning_rate: float
:param learning_rate: learning rate used in the finetune stage
:type pretraining_epochs: int
:param pretraining_epochs: number of epoch to do pretraining
:type pretrain_lr: float
:param pretrain_lr: learning rate to be used during pre-training
:type k: int
:param k: number of Gibbs steps in CD/PCD
:type training_epochs: int
:param training_epochs: maximal number of iterations ot run the optimizer
:type dataset: string
:param dataset: path the the pickled dataset
:type batch_size: int
:param batch_size: the size of a minibatch
"""
theano.config.blas.ldflags="-lopenblas"
datasets = load_data(dataset)
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_train_batches = train_set_x.get_value(borrow=True).shape[0] / batch_size
# numpy random generator
numpy_rng = numpy.random.RandomState(123)
print '... building the model'
# construct the Deep Belief Network
dbn = DBN(numpy_rng=numpy_rng, n_ins=28 * 28,
hidden_layers_sizes=[1000, 1000, 1000],
n_outs=10)
#########################
# PRETRAINING THE MODEL #
#########################
print '... getting the pretraining functions'
pretraining_fns = dbn.pretraining_functions(train_set_x=train_set_x,
batch_size=batch_size,
k=k)
print '... pre-training the model'
start_time = time.clock()
## Pre-train layer-wise
for i in xrange(dbn.n_layers):
# go through pretraining epochs
for epoch in xrange(pretraining_epochs):
# go through the training set
c = []
for batch_index in xrange(n_train_batches):
c.append(pretraining_fns[i](index=batch_index,
lr=pretrain_lr))
print 'Pre-training layer %i, epoch %d, cost ' % (i, epoch),
print numpy.mean(c)
end_time = time.clock()
print >> sys.stderr, ('The pretraining code for file ' +
os.path.split(__file__)[1] +
' ran for %.2fm' % ((end_time - start_time) / 60.))
########################
# FINETUNING THE MODEL #
########################
# get the training, validation and testing function for the model
print '... getting the finetuning functions'
train_fn, validate_model, test_model = dbn.build_finetune_functions(
datasets=datasets, batch_size=batch_size,
learning_rate=finetune_lr)
print '... finetunning the model'
# early-stopping parameters
patience = 4 * n_train_batches # look as this many examples regardless
patience_increase = 2. # wait this much longer when a new best is
# found
improvement_threshold = 0.995 # a relative improvement of this much is
# considered significant
validation_frequency = min(n_train_batches, patience / 2)
# go through this many
# minibatche before checking the network
# on the validation set; in this case we
# check every epoch
best_params = None
best_validation_loss = numpy.inf
test_score = 0.
start_time = time.clock()
done_looping = False
epoch = 0
while (epoch < training_epochs) and (not done_looping):
epoch = epoch + 1
for minibatch_index in xrange(n_train_batches):
minibatch_avg_cost = train_fn(minibatch_index)
iter = (epoch - 1) * n_train_batches + minibatch_index
if (iter + 1) % validation_frequency == 0:
validation_losses = validate_model()
this_validation_loss = numpy.mean(validation_losses)
print('epoch %i, minibatch %i/%i, validation error %f %%' % \
(epoch, minibatch_index + 1, n_train_batches,
this_validation_loss * 100.))
# if we got the best validation score until now
if this_validation_loss < best_validation_loss:
#improve patience if loss improvement is good enough
if (this_validation_loss < best_validation_loss *
improvement_threshold):
patience = max(patience, iter * patience_increase)
# save best validation score and iteration number
best_validation_loss = this_validation_loss
best_iter = iter
# test it on the test set
test_losses = test_model()
test_score = numpy.mean(test_losses)
print((' epoch %i, minibatch %i/%i, test error of '
'best model %f %%') %
(epoch, minibatch_index + 1, n_train_batches,
test_score * 100.))
if patience <= iter:
done_looping = True
break
end_time = time.clock()
print(('Optimization complete with best validation score of %f %%,'
'with test performance %f %%') %
(best_validation_loss * 100., test_score * 100.))
print >> sys.stderr, ('The fine tuning code for file ' +
os.path.split(__file__)[1] +
' ran for %.2fm' % ((end_time - start_time)
/ 60.))
def evaluate_lenet5(learning_rate=0.1, n_epochs=200,
dataset='../data/mnist.pkl.gz',
nkerns=[20, 50], batch_size=500):
""" Demonstrates lenet on MNIST dataset
:type learning_rate: float
:param learning_rate: learning rate used (factor for the stochastic
gradient)
:type n_epochs: int
:param n_epochs: maximal number of epochs to run the optimizer
:type dataset: string
:param dataset: path to the dataset used for training /testing (MNIST here)
:type nkerns: list of ints
:param nkerns: number of kernels on each layer
"""
rng = numpy.random.RandomState(23455)
datasets = load_data(dataset)
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_train_batches = train_set_x.get_value(borrow=True).shape[0]
n_valid_batches = valid_set_x.get_value(borrow=True).shape[0]
n_test_batches = test_set_x.get_value(borrow=True).shape[0]
n_train_batches /= batch_size
n_valid_batches /= batch_size
n_test_batches /= batch_size
# allocate symbolic variables for the data
index = T.lscalar() # index to a [mini]batch
x = T.matrix('x') # the data is presented as rasterized images
y = T.ivector('y') # the labels are presented as 1D vector of
# [int] labels
ishape = (28, 28) # this is the size of MNIST images
######################
# BUILD ACTUAL MODEL #
######################
print '... building the model'
# Reshape matrix of rasterized images of shape (batch_size,28*28)
# to a 4D tensor, compatible with our LeNetConvPoolLayer
layer0_input = x.reshape((batch_size, 1, 28, 28))
# Construct the first convolutional pooling layer:
# filtering reduces the image size to (28-5+1,28-5+1)=(24,24)
# maxpooling reduces this further to (24/2,24/2) = (12,12)
# 4D output tensor is thus of shape (batch_size,nkerns[0],12,12)
layer0 = LeNetConvPoolLayer(rng, input=layer0_input,
image_shape=(batch_size, 1, 28, 28),
filter_shape=(nkerns[0], 1, 5, 5), poolsize=(2, 2))
# Construct the second convolutional pooling layer
# filtering reduces the image size to (12-5+1,12-5+1)=(8,8)
# maxpooling reduces this further to (8/2,8/2) = (4,4)
# 4D output tensor is thus of shape (nkerns[0],nkerns[1],4,4)
layer1 = LeNetConvPoolLayer(rng, input=layer0.output,
image_shape=(batch_size, nkerns[0], 12, 12),
filter_shape=(nkerns[1], nkerns[0], 5, 5), poolsize=(2, 2))
# the TanhLayer being fully-connected, it operates on 2D matrices of
# shape (batch_size,num_pixels) (i.e matrix of rasterized images).
# This will generate a matrix of shape (20,32*4*4) = (20,512)
layer2_input = layer1.output.flatten(2)
# construct a fully-connected sigmoidal layer
layer2 = HiddenLayer(rng, input=layer2_input, n_in=nkerns[1] * 4 * 4,
n_out=500, activation=T.tanh)
# classify the values of the fully-connected sigmoidal layer
layer3 = LogisticRegression(input=layer2.output, n_in=500, n_out=10)
# the cost we minimize during training is the NLL of the model
cost = layer3.negative_log_likelihood(y)
# create a function to compute the mistakes that are made by the model
test_model = theano.function([index], layer3.errors(y),
givens={
x: test_set_x[index * batch_size: (index + 1) * batch_size],
y: test_set_y[index * batch_size: (index + 1) * batch_size]})
validate_model = theano.function([index], layer3.errors(y),
givens={
x: valid_set_x[index * batch_size: (index + 1) * batch_size],
y: valid_set_y[index * batch_size: (index + 1) * batch_size]})
# create a list of all model parameters to be fit by gradient descent
params = layer3.params + layer2.params + layer1.params + layer0.params
# create a list of gradients for all model parameters
grads = T.grad(cost, params)
# train_model is a function that updates the model parameters by
# SGD Since this model has many parameters, it would be tedious to
# manually create an update rule for each model parameter. We thus
# create the updates list by automatically looping over all
# (params[i],grads[i]) pairs.
updates = []
for param_i, grad_i in zip(params, grads):
updates.append((param_i, param_i - learning_rate * grad_i))
train_model = theano.function([index], cost, updates=updates,
givens={
x: train_set_x[index * batch_size: (index + 1) * batch_size],
y: train_set_y[index * batch_size: (index + 1) * batch_size]})
###############
# TRAIN MODEL #
###############
print '... training'
# early-stopping parameters
patience = 10000 # look as this many examples regardless
patience_increase = 2 # wait this much longer when a new best is
# found
improvement_threshold = 0.995 # a relative improvement of this much is
# considered significant
validation_frequency = min(n_train_batches, patience / 2)
# go through this many
# minibatche before checking the network
# on the validation set; in this case we
# check every epoch
best_params = None
best_validation_loss = numpy.inf
best_iter = 0
test_score = 0.
start_time = time.clock()
epoch = 0
done_looping = False
while (epoch < n_epochs) and (not done_looping):
epoch = epoch + 1
for minibatch_index in xrange(n_train_batches):
iter = (epoch - 1) * n_train_batches + minibatch_index
if iter % 100 == 0:
print 'training @ iter = ', iter
cost_ij = train_model(minibatch_index)
if (iter + 1) % validation_frequency == 0:
# compute zero-one loss on validation set
validation_losses = [validate_model(i) for i
in xrange(n_valid_batches)]
this_validation_loss = numpy.mean(validation_losses)
print('epoch %i, minibatch %i/%i, validation error %f %%' % \
(epoch, minibatch_index + 1, n_train_batches, \
this_validation_loss * 100.))
# if we got the best validation score until now
if this_validation_loss < best_validation_loss:
#improve patience if loss improvement is good enough
if this_validation_loss < best_validation_loss * \
improvement_threshold:
patience = max(patience, iter * patience_increase)
# save best validation score and iteration number
best_validation_loss = this_validation_loss
best_iter = iter
# test it on the test set
test_losses = [test_model(i) for i in xrange(n_test_batches)]
test_score = numpy.mean(test_losses)
print((' epoch %i, minibatch %i/%i, test error of best '
'model %f %%') %
(epoch, minibatch_index + 1, n_train_batches,
test_score * 100.))
if patience <= iter:
done_looping = True
break
end_time = time.clock()
print('Optimization complete.')
print('Best validation score of %f %% obtained at iteration %i,'\
'with test performance %f %%' %
(best_validation_loss * 100., best_iter + 1, test_score * 100.))
print >> sys.stderr, ('The code for file ' +
os.path.split(__file__)[1] +
' ran for %.2fm' % ((end_time - start_time) / 60.))
def test_mlp(learning_rate=0.1, L1_reg=0.00, L2_reg=0.0001, n_epochs=1000,
dataset='../data/old.murray.all.pkl.gz', batch_size=1000, n_hidden=500,
n_in=180, n_out=2, hidden_layers_sizes=[500,500], outfile="ami-mlp-eval.pkl"):
"""
Demonstrate stochastic gradient descent optimization for a multilayer
perceptron
:type learning_rate: float
:param learning_rate: learning rate used (factor for the stochastic
gradient
:type L1_reg: float
:param L1_reg: L1-norm's weight when added to the cost (see
regularization)
:type L2_reg: float
:param L2_reg: L2-norm's weight when added to the cost (see
regularization)
:type n_epochs: int
:param n_epochs: maximal number of epochs to run the optimizer
:type dataset: string
:param dataset: the path of the MNIST dataset file from
http://www.iro.umontreal.ca/~lisa/deep/data/mnist/mnist.pkl.gz
"""
datasets = load_data(dataset)
train_set_x, train_set_y, train_set_z = datasets[0]
valid_set_x, valid_set_y, valid_set_z = datasets[1]
test_set_x, test_set_y, test_set_z = datasets[2]
# compute number of minibatches for training, validation and testing
n_train_batches = train_set_x.get_value(borrow=True).shape[0] / batch_size
n_valid_batches = valid_set_x.get_value(borrow=True).shape[0] / batch_size
n_test_batches = test_set_x.get_value(borrow=True).shape[0] / batch_size
n_test_size = test_set_x.get_value(borrow=True).shape[0]
n_dev_size = valid_set_x.get_value(borrow=True).shape[0]
n_train_size = train_set_x.get_value(borrow=True).shape[0]
print n_train_size
######################
# BUILD ACTUAL MODEL #
######################
print '... building the model'
# allocate symbolic variables for the data
index = T.lscalar() # index to a [mini]batch
x = T.matrix('x') # the data is presented as rasterized images
y = T.ivector('y') # the labels are presented as 1D vector of
# [int] labels
z = T.ivector('z')
rng = numpy.random.RandomState(1234)
# construct the MLP class
classifier = MLP(rng=rng, input=x, n_in=n_in,
n_hidden=n_hidden, n_out=n_out, hidden_layers_sizes=hidden_layers_sizes)
# the cost we minimize during training is the negative log likelihood of
# the model plus the regularization terms (L1 and L2); cost is expressed
# here symbolically
cost = classifier.negative_log_likelihood(y) \
+ L1_reg * classifier.L1 \
+ L2_reg * classifier.L2_sqr
# compiling a Theano function that computes the mistakes that are made
# by the model on a minibatch
test_model = theano.function(inputs=[index],
outputs=classifier.errors(y),
givens={
x: test_set_x[index * batch_size:(index + 1) * batch_size],
y: test_set_y[index * batch_size:(index + 1) * batch_size]})
validate_model = theano.function(inputs=[index],
outputs=classifier.errors(y),
givens={
x: valid_set_x[index * batch_size:(index + 1) * batch_size],
y: valid_set_y[index * batch_size:(index + 1) * batch_size]})
pred_prob_model = theano.function(inputs=[index],
outputs=[y, classifier.get_y_pred(y), classifier.get_p_y_given_x(y), z],
givens={
x: test_set_x[index: (index + 1)],
y: test_set_y[index: (index + 1)],
z: test_set_z[index: (index + 1)]})
pred_prob_dev = theano.function(inputs=[index],
outputs=[y, classifier.get_y_pred(y), classifier.get_p_y_given_x(y), z],
givens={
x: valid_set_x[index: (index + 1)],
y: valid_set_y[index: (index + 1)],
z: valid_set_z[index: (index + 1)]})
pred_prob_train = theano.function(inputs=[index],
outputs=[y, classifier.get_y_pred(y), classifier.get_p_y_given_x(y), z],
givens={
x: train_set_x[index: (index + 1)],
y: train_set_y[index: (index + 1)],
z: train_set_z[index: (index + 1)]})
# the resulting gradients will be stored in a list gparams
gparams = []
for param in classifier.params:
gparam = T.grad(cost, param)
gparams.append(gparam)
# specify how to update the parameters of the model as a list of
# (variable, update expression) pairs
updates = []
# given two list the zip A = [a1, a2, a3, a4] and B = [b1, b2, b3, b4] of
# same length, zip generates a list C of same size, where each element
# is a pair formed from the two lists :
# C = [(a1, b1), (a2, b2), (a3, b3), (a4, b4)]
for param, gparam in zip(classifier.params, gparams):
updates.append((param, param - learning_rate * gparam))
# compiling a Theano function `train_model` that returns the cost, but
# in the same time updates the parameter of the model based on the rules
# defined in `updates`
train_model = theano.function(inputs=[index], outputs=cost,
updates=updates,
givens={
x: train_set_x[index * batch_size:(index + 1) * batch_size],
y: train_set_y[index * batch_size:(index + 1) * batch_size]})
###############
# TRAIN MODEL #
###############
print '... training'
# early-stopping parameters
patience = 10000 # look as this many examples regardless
patience_increase = 2 # wait this much longer when a new best is
# found
improvement_threshold = 0.995 # a relative improvement of this much is
# considered significant
validation_frequency = min(n_train_batches, patience / 2)
# go through this many
# minibatche before checking the network
# on the validation set; in this case we
# check every epoch
best_params = None
best_validation_loss = numpy.inf
best_iter = 0
test_score = 0.
start_time = time.clock()
epoch = 0
done_looping = False
while (epoch < n_epochs) and (not done_looping):
epoch = epoch + 1
for minibatch_index in xrange(n_train_batches):
minibatch_avg_cost = train_model(minibatch_index)
# iteration number
iter = (epoch - 1) * n_train_batches + minibatch_index
if (iter + 1) % validation_frequency == 0:
# compute zero-one loss on validation set
validation_losses = [validate_model(i) for i
in xrange(n_valid_batches)]
this_validation_loss = numpy.mean(validation_losses)
print('epoch %i, minibatch %i/%i, validation error %f %%' %
(epoch, minibatch_index + 1, n_train_batches,
this_validation_loss * 100.))
# if we got the best validation score until now
if this_validation_loss < best_validation_loss:
#improve patience if loss improvement is good enough
if this_validation_loss < best_validation_loss * \
improvement_threshold:
patience = max(patience, iter * patience_increase)
best_validation_loss = this_validation_loss
best_iter = iter
# test it on the test set
test_losses = [test_model(i) for i
in xrange(n_test_batches)]
test_score = numpy.mean(test_losses)
print((' epoch %i, minibatch %i/%i, test error of '
'best model %f %%') %
(epoch, minibatch_index + 1, n_train_batches,
test_score * 100.))
if patience <= iter:
done_looping = True
break
end_time = time.clock()
print "End training"
print "Get probabilities for all data"
print "train"
train_preds = [pred_prob_train(i) for i in xrange(n_train_size)]
print "test"
test_preds = [pred_prob_model(i) for i in xrange(n_test_size)]
print "dev"
dev_preds = [pred_prob_dev(i) for i in xrange(n_dev_size)]
print len(train_preds), len(test_preds), len(dev_preds)
print "Dump probs"
save_file = open(outfile, 'wb')
cPickle.dump([map(lambda y: [y[0][0], y[1][0], y[2][0][1], y[3][0]], dev_preds),
map(lambda y: [y[0][0], y[1][0], y[2][0][1], y[3][0]], test_preds),
map(lambda y: [y[0][0], y[1][0], y[2][0][1], y[3][0]], train_preds)], save_file, -1)
print "Dump params"
paramfile = "params-" + outfile
print paramfile
save_file = open(paramfile, 'wb')
currparams = classifier.get_params()
cPickle.dump(currparams, save_file, -1)
print(('Optimization complete. Best validation score of %f %% '
'obtained at iteration %i, with test performance %f %%') %
(best_validation_loss * 100., best_iter + 1, test_score * 100.))
print >> sys.stderr, ('The code for file ' +
os.path.split(__file__)[1] +
' ran for %.2fm' % ((end_time - start_time) / 60.))
def test_rbm(learning_rate=0.1,
training_epochs=15,
dataset='../data/mnist.pkl.gz',
batch_size=20,
n_chains=20,
n_samples=10,
output_folder='rbm_plots',
n_hidden=500):
"""
Demonstrate how to train and afterwards sample from it using Theano.
This is demonstrated on MNIST.
:param learning_rate: learning rate used for training the RBM
:param training_epochs: number of epochs used for training
:param dataset: path the the pickled dataset
:param batch_size: size of a batch used to train the RBM
:param n_chains: number of parallel Gibbs chains to be used for sampling
:param n_samples: number of samples to plot for each chain
"""
datasets = load_data(dataset)
train_set_x, train_set_y = datasets[0]
test_set_x, test_set_y = datasets[2]
# compute number of minibatches for training, validation and testing
n_train_batches = train_set_x.get_value(borrow=True).shape[0] / batch_size
# allocate symbolic variables for the data
index = T.lscalar() # index to a [mini]batch
x = T.matrix('x') # the data is presented as rasterized images
rng = numpy.random.RandomState(123)
theano_rng = RandomStreams(rng.randint(2**30))
# initialize storage for the persistent chain (state = hidden
# layer of chain)
persistent_chain = theano.shared(numpy.zeros((batch_size, n_hidden),
dtype=theano.config.floatX),
borrow=True)
# construct the RBM class
rbm = RBM(input=x,
n_visible=28 * 28,
n_hidden=n_hidden,
numpy_rng=rng,
theano_rng=theano_rng)
# get the cost and the gradient corresponding to one step of CD-15
cost, updates = rbm.get_cost_updates(lr=learning_rate,
persistent=persistent_chain,
k=15)
#################################
# Training the RBM #
#################################
if not os.path.isdir(output_folder):
os.makedirs(output_folder)
os.chdir(output_folder)
# it is ok for a theano function to have no output
# the purpose of train_rbm is solely to update the RBM parameters
train_rbm = theano.function(
[index],
cost,
updates=updates,
givens={x: train_set_x[index * batch_size:(index + 1) * batch_size]},
name='train_rbm')
plotting_time = 0.
start_time = time.clock()
# go through training epochs
for epoch in xrange(training_epochs):
# go through the training set
mean_cost = []
for batch_index in xrange(n_train_batches):
mean_cost += [train_rbm(batch_index)]
print 'Training epoch %d, cost is ' % epoch, numpy.mean(mean_cost)
# Plot filters after each training epoch
plotting_start = time.clock()
# Construct image from the weight matrix
image = PIL.Image.fromarray(
tile_raster_images(X=rbm.W.get_value(borrow=True).T,
img_shape=(28, 28),
tile_shape=(10, 10),
tile_spacing=(1, 1)))
image.save('filters_at_epoch_%i.png' % epoch)
plotting_stop = time.clock()
plotting_time += (plotting_stop - plotting_start)
end_time = time.clock()
pretraining_time = (end_time - start_time) - plotting_time
print('Training took %f minutes' % (pretraining_time / 60.))
#################################
# Sampling from the RBM #
#################################
# find out the number of test samples
number_of_test_samples = test_set_x.get_value(borrow=True).shape[0]
# pick random test examples, with which to initialize the persistent chain
test_idx = rng.randint(number_of_test_samples - n_chains)
persistent_vis_chain = theano.shared(
numpy.asarray(test_set_x.get_value(borrow=True)[test_idx:test_idx +
n_chains],
dtype=theano.config.floatX))
plot_every = 1000
# define one step of Gibbs sampling (mf = mean-field) define a
# function that does `plot_every` steps before returning the
# sample for plotting
[presig_hids, hid_mfs, hid_samples, presig_vis,
vis_mfs, vis_samples], updates = \
theano.scan(rbm.gibbs_vhv,
outputs_info=[None, None, None, None,
None, persistent_vis_chain],
n_steps=plot_every)
# add to updates the shared variable that takes care of our persistent
# chain :.
updates.update({persistent_vis_chain: vis_samples[-1]})
# construct the function that implements our persistent chain.
# we generate the "mean field" activations for plotting and the actual
# samples for reinitializing the state of our persistent chain
sample_fn = theano.function([], [vis_mfs[-1], vis_samples[-1]],
updates=updates,
name='sample_fn')
# create a space to store the image for plotting ( we need to leave
# room for the tile_spacing as well)
image_data = numpy.zeros((29 * n_samples + 1, 29 * n_chains - 1),
dtype='uint8')
for idx in xrange(n_samples):
# generate `plot_every` intermediate samples that we discard,
# because successive samples in the chain are too correlated
vis_mf, vis_sample = sample_fn()
print ' ... plotting sample ', idx
image_data[29 * idx:29 * idx + 28, :] = tile_raster_images(
X=vis_mf,
img_shape=(28, 28),
tile_shape=(1, n_chains),
tile_spacing=(1, 1))
# construct image
image = PIL.Image.fromarray(image_data)
image.save('samples.png')
os.chdir('../')
def test_mlp(learning_rate = 0.01, L1_reg = 0.00,L2_reg = 0.0001,n_epochs=1000,dataset = 'mnist.pkl.gz',batch_size=20,n_hidden = 500):
datasets = load_data(dataset)
train_set_x ,train_set_y = datasets[0]
valid_set_x,valid_set_y = datasets[1]
test_set_x, test_set_y = datasets[2]
n_train_batches = train_set_x.get_value(borrow=True).shape[0] / batch_size
n_valid_batches = valid_set_x.get_value(borrow=True).shape[0]/batch_size
n_test_batches = test_set_x.get_value(borrow=True).shape[0]/batch_size
print 'building the model'
index = T.lscalar()
x = T.matrix('x')
y = T.ivector('y')
rng = np.random.RandomState()
classifier = MLP(
rng = rng,
input=x,
n_in = 28*28,
n_hidden = n_hidden,
n_out = 10
)
cost = (
classifier.negative_log_likelihood(y)
+L1_reg*classifier.L1
+L2_reg*classifier.L2_sqr
)
test_model = theano.function(
inputs=[index],
outputs=classifier.errors(y),
givens={
x:test_set_x[index *batch_size:(index+1)*batch_size],
y:test_set_y[index *batch_size:(index+1)*batch_size]
}
)
validate_model = theano.function(
inputs=[index],
outputs=classifier.errors(y),
givens={
x: valid_set_x[index * batch_size:(index + 1) * batch_size],
y: valid_set_y[index * batch_size:(index + 1) * batch_size]
}
)
gparams = [T.grad(cost,param) for param in classifier.params]
updates=[
(param,param-learning_rate * gparam)
for param,gparam in zip(classifier.params,gparams)
]
train_model = theano.function(
inputs = [index],
outputs= cost,
updates = updates,
givens={
x:train_set_x[index*batch_size:(index+1)*batch_size],
y:train_set_y[index*batch_size:(index+1)*batch_size]
}
)
print 'training'
patience = 10000
patience_incresase = 2
improvement_threshold = 0.995
validation_frequency = min(n_train_batches,patience/2)
best_validation_loss = np.inf
best_iter = 0
test_score = 0.
start_time = timeit.default_timer()
epoch = 0
done_looping = False
while(epoch < n_epochs) and (not done_looping):
epoch = epoch + 1
for minibatch_index in xrange(n_train_batches):
minibatch_avg_cost = train_model(minibatch_index)
iter = (epoch -1) * n_train_batches + minibatch_index
if (iter + 1) % validation_frequency == 0:
validation_losses = [validate_model(i) for i in xrange(n_valid_batches)]
this_validation_loss = np.mean(validation_losses)
print (
'epoch %i , minibatch %i / %i,validation error %f %%' %
(
epoch,
minibatch_index + 1,
n_train_batches,
this_validation_loss * 100
)
)
if this_validation_loss < best_validation_loss:
if(
this_validation_loss < best_validation_loss * improvement_threshold
):
patience = max(patience, iter*patience_incresase)
best_validation_loss = this_validation_loss
best_iter = iter
test_losses = [test_model(i) for i in xrange(n_test_batches)]
test_score = np.mean(test_losses)
print ((' epoch %i,minibatch %i/%i, test error of best model %f %%')%
(
epoch,minibatch_index+1,n_train_batches,test_score *100.
))
if patience <= iter:
done_looping = True
break
end_time = timeit.default_timer()
print(('Optimization complete. Best validation score of %f %% '
'obtained at iteration %i, with test performance %f %%') %
(best_validation_loss * 100., best_iter + 1, test_score * 100.))
print >> sys.stderr, ('The code for file ' +
os.path.split(__file__)[1] +
' ran for %.2fm' % ((end_time - start_time) / 60.))
def test_dA(learning_rate=0.1,
training_epochs=15,
dataset='mnist.pkl.gz',
batch_size=20,
output_folder='dA_plots'):
datasets = load_data(dataset)
train_set_x, train_set_y = datasets[0]
n_train_batches = train_set_x.get_value(borrow=True).shape[0] / batch_size
index = T.lscalar() # index to a [mini]batch
x = T.matrix('x') # the data is presented as rasterized images
if not os.path.isdir(output_folder):
os.makedirs(output_folder)
os.chdir(output_folder)
####################################
# BUILDING THE MODEL NO CORRUPTION #
####################################
rng = numpy.random.RandomState(123)
theano_rng = RandomStreams(rng.randint(2**30))
da = dA(numpy_rng=rng,
theano_rng=theano_rng,
input=x,
n_visible=28 * 28,
n_hidden=500)
cost, updates = da.get_cost_updates(corruption_level=0.,
learning_rate=learning_rate)
train_da = theano.function(
[index],
cost,
updates=updates,
givens={x: train_set_x[index * batch_size:(index + 1) * batch_size]})
start_time = timeit.default_timer()
############
# TRAINING #
############
# go through training epochs
for epoch in xrange(training_epochs):
# go through trainng set
c = []
for batch_index in xrange(n_train_batches):
c.append(train_da(batch_index))
print 'Training epoch %d, cost ' % epoch, numpy.mean(c)
end_time = timeit.default_timer()
training_time = (end_time - start_time)
print >> sys.stderr, ('The no corruption code for file ' +
os.path.split(__file__)[1] + ' ran for %.2fm' %
((training_time) / 60.))
'''
image = Image.fromarray(
tile_raster_images(X=da.W.get_value(borrow=True).T,
img_shape=(28, 28), tile_shape=(10, 10),
tile_spacing=(1, 1)))
image.save('filters_corruption_0.png')
'''
# start-snippet-3
#####################################
# BUILDING THE MODEL CORRUPTION 30% #
#####################################
rng = numpy.random.RandomState(123)
theano_rng = RandomStreams(rng.randint(2**30))
da = dA(numpy_rng=rng,
theano_rng=theano_rng,
input=x,
n_visible=28 * 28,
n_hidden=500)
cost, updates = da.get_cost_updates(corruption_level=0.3,
learning_rate=learning_rate)
train_da = theano.function(
[index],
cost,
updates=updates,
givens={x: train_set_x[index * batch_size:(index + 1) * batch_size]})
start_time = timeit.default_timer()
############
# TRAINING #
############
# go through training epochs
for epoch in xrange(training_epochs):
# go through trainng set
c = []
for batch_index in xrange(n_train_batches):
c.append(train_da(batch_index))
print 'Training epoch %d, cost ' % epoch, numpy.mean(c)
end_time = timeit.default_timer()
training_time = (end_time - start_time)
print >> sys.stderr, ('The 30% corruption code for file ' +
os.path.split(__file__)[1] + ' ran for %.2fm' %
(training_time / 60.))
borrow=True
)
#b_values = np.zeros((n_out,), dtype=theano.config.floatX)
#b = theano.shared(value=b_values, name='b', borrow=True)
self.w = w
self.b = b
self.output = T.tanh(T.dot(input,self.w)+self.b)
self.params = [self.w,self.b]
# In[5]:
datasets = load_data('mnist.pkl.gz')
batch_size = 20
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_train_batches = train_set_x.get_value(borrow=True).shape[0] / batch_size
n_valid_batches = valid_set_x.get_value(borrow=True).shape[0] / batch_size
n_test_batches = test_set_x.get_value(borrow=True).shape[0] / batch_size
# In[6]:
dtype=theano.config.floatX),
name='b',
borrow=True)
#b_values = np.zeros((n_out,), dtype=theano.config.floatX)
#b = theano.shared(value=b_values, name='b', borrow=True)
self.w = w
self.b = b
self.output = T.tanh(T.dot(input, self.w) + self.b)
self.params = [self.w, self.b]
# In[5]:
datasets = load_data('mnist.pkl.gz')
batch_size = 20
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_train_batches = train_set_x.get_value(borrow=True).shape[0] / batch_size
n_valid_batches = valid_set_x.get_value(borrow=True).shape[0] / batch_size
n_test_batches = test_set_x.get_value(borrow=True).shape[0] / batch_size
# In[6]:
### : Testing the hidden layer ###
def test_mlp(learning_rate=0.01,
L1_reg=0.00,
L2_reg=0.0001,
n_epochs=1000,
dataset='../data/mnist.pkl.gz',
batch_size=20,
n_hidden=500):
"""
Demonstrate stochastic gradient descent optimization for a multilayer
perceptron
This is demonstrated on MNIST.
:type learning_rate: float
:param learning_rate: learning rate used (factor for the stochastic
gradient
:type L1_reg: float
:param L1_reg: L1-norm's weight when added to the cost (see
regularization)
:type L2_reg: float
:param L2_reg: L2-norm's weight when added to the cost (see
regularization)
:type n_epochs: int
:param n_epochs: maximal number of epochs to run the optimizer
:type dataset: string
:param dataset: the path of the MNIST dataset file from
http://www.iro.umontreal.ca/~lisa/deep/data/mnist/mnist.pkl.gz
"""
datasets = load_data(dataset)
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_train_batches = train_set_x.get_value(borrow=True).shape[0] / batch_size
n_valid_batches = valid_set_x.get_value(borrow=True).shape[0] / batch_size
n_test_batches = test_set_x.get_value(borrow=True).shape[0] / batch_size
######################
# BUILD ACTUAL MODEL #
######################
print '... building the model'
# allocate symbolic variables for the data
index = T.lscalar() # index to a [mini]batch
x = T.matrix('x') # the data is presented as rasterized images
y = T.ivector('y') # the labels are presented as 1D vector of
# [int] labels
rng = numpy.random.RandomState(1234)
# construct the MLP class
classifier = MLP(rng=rng,
input=x,
n_in=28 * 28,
n_hidden=n_hidden,
n_out=10)
# the cost we minimize during training is the negative log likelihood of
# the model plus the regularization terms (L1 and L2); cost is expressed
# here symbolically
cost = classifier.negative_log_likelihood(y) \
+ L1_reg * classifier.L1 \
+ L2_reg * classifier.L2_sqr
# compiling a Theano function that computes the mistakes that are made
# by the model on a minibatch
test_model = theano.function(
inputs=[index],
outputs=classifier.errors(y),
givens={
x: test_set_x[index * batch_size:(index + 1) * batch_size],
y: test_set_y[index * batch_size:(index + 1) * batch_size]
})
validate_model = theano.function(
inputs=[index],
outputs=classifier.errors(y),
givens={
x: valid_set_x[index * batch_size:(index + 1) * batch_size],
y: valid_set_y[index * batch_size:(index + 1) * batch_size]
})
# compute the gradient of cost with respect to theta (sotred in params)
# the resulting gradients will be stored in a list gparams
gparams = []
for param in classifier.params:
gparam = T.grad(cost, param)
gparams.append(gparam)
# specify how to update the parameters of the model as a list of
# (variable, update expression) pairs
updates = []
# given two list the zip A = [a1, a2, a3, a4] and B = [b1, b2, b3, b4] of
# same length, zip generates a list C of same size, where each element
# is a pair formed from the two lists :
# C = [(a1, b1), (a2, b2), (a3, b3), (a4, b4)]
for param, gparam in zip(classifier.params, gparams):
updates.append((param, param - learning_rate * gparam))
# compiling a Theano function `train_model` that returns the cost, but
# in the same time updates the parameter of the model based on the rules
# defined in `updates`
train_model = theano.function(
inputs=[index],
outputs=cost,
updates=updates,
givens={
x: train_set_x[index * batch_size:(index + 1) * batch_size],
y: train_set_y[index * batch_size:(index + 1) * batch_size]
})
###############
# TRAIN MODEL #
###############
print '... training'
# early-stopping parameters
patience = 10000 # look as this many examples regardless
patience_increase = 2 # wait this much longer when a new best is
# found
improvement_threshold = 0.995 # a relative improvement of this much is
# considered significant
validation_frequency = min(n_train_batches, patience / 2)
# go through this many
# minibatche before checking the network
# on the validation set; in this case we
# check every epoch
best_params = None
best_validation_loss = numpy.inf
best_iter = 0
test_score = 0.
start_time = time.clock()
epoch = 0
done_looping = False
while (epoch < n_epochs) and (not done_looping):
epoch = epoch + 1
for minibatch_index in xrange(n_train_batches):
minibatch_avg_cost = train_model(minibatch_index)
# iteration number
iter = (epoch - 1) * n_train_batches + minibatch_index
if (iter + 1) % validation_frequency == 0:
# compute zero-one loss on validation set
validation_losses = [
validate_model(i) for i in xrange(n_valid_batches)
]
this_validation_loss = numpy.mean(validation_losses)
print('epoch %i, minibatch %i/%i, validation error %f %%' %
(epoch, minibatch_index + 1, n_train_batches,
this_validation_loss * 100.))
# if we got the best validation score until now
if this_validation_loss < best_validation_loss:
#improve patience if loss improvement is good enough
if this_validation_loss < best_validation_loss * \
improvement_threshold:
patience = max(patience, iter * patience_increase)
best_validation_loss = this_validation_loss
best_iter = iter
# test it on the test set
test_losses = [
test_model(i) for i in xrange(n_test_batches)
]
test_score = numpy.mean(test_losses)
print((' epoch %i, minibatch %i/%i, test error of '
'best model %f %%') %
(epoch, minibatch_index + 1, n_train_batches,
test_score * 100.))
if patience <= iter:
done_looping = True
break
end_time = time.clock()
print(('Optimization complete. Best validation score of %f %% '
'obtained at iteration %i, with test performance %f %%') %
(best_validation_loss * 100., best_iter + 1, test_score * 100.))
print >> sys.stderr, ('The code for file ' + os.path.split(__file__)[1] +
' ran for %.2fm' % ((end_time - start_time) / 60.))
def evaluate_lenet5(learning_rate=0.1,
n_epochs=200,
dataset='../data/mnist.pkl.gz',
nkerns=[20, 50],
batch_size=500):
""" Demonstrates lenet on MNIST dataset
:type learning_rate: float
:param learning_rate: learning rate used (factor for the stochastic
gradient)
:type n_epochs: int
:param n_epochs: maximal number of epochs to run the optimizer
:type dataset: string
:param dataset: path to the dataset used for training /testing (MNIST here)
:type nkerns: list of ints
:param nkerns: number of kernels on each layer
"""
rng = numpy.random.RandomState(23455)
datasets = load_data(dataset)
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_train_batches = train_set_x.get_value(borrow=True).shape[0]
n_valid_batches = valid_set_x.get_value(borrow=True).shape[0]
n_test_batches = test_set_x.get_value(borrow=True).shape[0]
n_train_batches /= batch_size
n_valid_batches /= batch_size
n_test_batches /= batch_size
# allocate symbolic variables for the data
index = T.lscalar() # index to a [mini]batch
x = T.matrix('x') # the data is presented as rasterized images
y = T.ivector('y') # the labels are presented as 1D vector of
# [int] labels
ishape = (28, 28) # this is the size of MNIST images
######################
# BUILD ACTUAL MODEL #
######################
print '... building the model'
# Reshape matrix of rasterized images of shape (batch_size,28*28)
# to a 4D tensor, compatible with our LeNetConvPoolLayer
layer0_input = x.reshape((batch_size, 1, 28, 28))
# Construct the first convolutional pooling layer:
# filtering reduces the image size to (28-5+1,28-5+1)=(24,24)
# maxpooling reduces this further to (24/2,24/2) = (12,12)
# 4D output tensor is thus of shape (batch_size,nkerns[0],12,12)
layer0 = LeNetConvPoolLayer(rng,
input=layer0_input,
image_shape=(batch_size, 1, 28, 28),
filter_shape=(nkerns[0], 1, 5, 5),
poolsize=(2, 2))
# Construct the second convolutional pooling layer
# filtering reduces the image size to (12-5+1,12-5+1)=(8,8)
# maxpooling reduces this further to (8/2,8/2) = (4,4)
# 4D output tensor is thus of shape (nkerns[0],nkerns[1],4,4)
layer1 = LeNetConvPoolLayer(rng,
input=layer0.output,
image_shape=(batch_size, nkerns[0], 12, 12),
filter_shape=(nkerns[1], nkerns[0], 5, 5),
poolsize=(2, 2))
# the TanhLayer being fully-connected, it operates on 2D matrices of
# shape (batch_size,num_pixels) (i.e matrix of rasterized images).
# This will generate a matrix of shape (20,32*4*4) = (20,512)
layer2_input = layer1.output.flatten(2)
# construct a fully-connected sigmoidal layer
layer2 = HiddenLayer(rng,
input=layer2_input,
n_in=nkerns[1] * 4 * 4,
n_out=500,
activation=T.tanh)
# classify the values of the fully-connected sigmoidal layer
layer3 = LogisticRegression(input=layer2.output, n_in=500, n_out=10)
# the cost we minimize during training is the NLL of the model
cost = layer3.negative_log_likelihood(y)
# create a function to compute the mistakes that are made by the model
test_model = theano.function(
[index],
layer3.errors(y),
givens={
x: test_set_x[index * batch_size:(index + 1) * batch_size],
y: test_set_y[index * batch_size:(index + 1) * batch_size]
})
validate_model = theano.function(
[index],
layer3.errors(y),
givens={
x: valid_set_x[index * batch_size:(index + 1) * batch_size],
y: valid_set_y[index * batch_size:(index + 1) * batch_size]
})
# create a list of all model parameters to be fit by gradient descent
params = layer3.params + layer2.params + layer1.params + layer0.params
# create a list of gradients for all model parameters
grads = T.grad(cost, params)
# train_model is a function that updates the model parameters by
# SGD Since this model has many parameters, it would be tedious to
# manually create an update rule for each model parameter. We thus
# create the updates list by automatically looping over all
# (params[i],grads[i]) pairs.
updates = []
for param_i, grad_i in zip(params, grads):
updates.append((param_i, param_i - learning_rate * grad_i))
train_model = theano.function(
[index],
cost,
updates=updates,
givens={
x: train_set_x[index * batch_size:(index + 1) * batch_size],
y: train_set_y[index * batch_size:(index + 1) * batch_size]
})
###############
# TRAIN MODEL #
###############
print '... training'
# early-stopping parameters
patience = 10000 # look as this many examples regardless
patience_increase = 2 # wait this much longer when a new best is
# found
improvement_threshold = 0.995 # a relative improvement of this much is
# considered significant
validation_frequency = min(n_train_batches, patience / 2)
# go through this many
# minibatche before checking the network
# on the validation set; in this case we
# check every epoch
best_params = None
best_validation_loss = numpy.inf
best_iter = 0
test_score = 0.
start_time = time.clock()
epoch = 0
done_looping = False
while (epoch < n_epochs) and (not done_looping):
epoch = epoch + 1
for minibatch_index in xrange(n_train_batches):
iter = (epoch - 1) * n_train_batches + minibatch_index
if iter % 100 == 0:
print 'training @ iter = ', iter
cost_ij = train_model(minibatch_index)
if (iter + 1) % validation_frequency == 0:
# compute zero-one loss on validation set
validation_losses = [
validate_model(i) for i in xrange(n_valid_batches)
]
this_validation_loss = numpy.mean(validation_losses)
print('epoch %i, minibatch %i/%i, validation error %f %%' % \
(epoch, minibatch_index + 1, n_train_batches, \
this_validation_loss * 100.))
# if we got the best validation score until now
if this_validation_loss < best_validation_loss:
#improve patience if loss improvement is good enough
if this_validation_loss < best_validation_loss * \
improvement_threshold:
patience = max(patience, iter * patience_increase)
# save best validation score and iteration number
best_validation_loss = this_validation_loss
best_iter = iter
# test it on the test set
test_losses = [
test_model(i) for i in xrange(n_test_batches)
]
test_score = numpy.mean(test_losses)
print(
(' epoch %i, minibatch %i/%i, test error of best '
'model %f %%') % (epoch, minibatch_index + 1,
n_train_batches, test_score * 100.))
if patience <= iter:
done_looping = True
break
end_time = time.clock()
print('Optimization complete.')
print('Best validation score of %f %% obtained at iteration %i,'\
'with test performance %f %%' %
(best_validation_loss * 100., best_iter + 1, test_score * 100.))
print >> sys.stderr, ('The code for file ' + os.path.split(__file__)[1] +
' ran for %.2fm' % ((end_time - start_time) / 60.))
def test_SdA(finetune_lr=0.1, pretraining_epochs=15,
pretraining_lr=0.001, training_epochs=1000,
dataset='mnist.pkl.gz', batch_size=1):
print('Loading Data'+'.'*20)
datasets =load_data(dataset)
trainSetX, trainSetY = datasets[0]
validSetX, validSetY = datasets[1]
testSetX, testSetY = datasets[2]
n_train_batches = trainSetX.get_value(borrow=True).shape[0] // batch_size
print('Building Model'+'.'*20)
numpy_rng = numpy.random.RandomState(1234)
sda = SdA(
numpy_rng=numpy_rng,
n_in=28*28,
hidden_layers_sizes=[1000,1000,1000],
n_out=10
)
#
# Pretraining the Model
#
print('.'*10+'getting the pretraining functions'+'.'*10)
pretraining_fns = sda.preTraining_functions(trainSetX=trainSetX,batch_size=batch_size)
print('.'*10+'pretraining the model'+'.'*10)
start_time = time.clock()
corruption_levels = [.1, .2, .3]
# pretraining layer-wise
for i in range(sda.n_layers):
# go through pretraining epochs
for epoch in range(pretraining_epochs):
# go through training set
c = []
for batch_index in range(n_train_batches):
c.append(pretraining_fns[i](index=batch_index,
corruption=corruption_levels[i],
lr=pretraining_lr))
print('Pretraining layer %i, epoch %d, cost ' %(i, epoch), numpy.mean(c))
end_time = time.clock()
print('The pretraining code for file '+
os.path.split(__file__)[1] +
' ran for %.2s' % (end_time - start_time))
#
# Finetuning the model
#
# get the training, validation, testing function for the model
print('.'*10+'getting the finetune functions'+','*20)
train_fn, valid_model, test_model = sda.build_finetune_functions(
datasets=datasets,
batch_size=batch_size,
learning_rate=finetune_lr
)
print('.'*20+'finetuning the model'+'.'*20)
# early stop parameter
patience = 10 * n_train_batches # look as this many examples regradless
patience_increase = 2. # wait this much longer when a new best is found
improvement_threshold = 0.95 # a relative improvement of this
# much is consider significant
validation_frequency = min(n_train_batches, patience/2)
# go through this many minibatch
# before checking the network
# on the validation set
best_validation_loss = numpy.inf
test_score = 0.
start_time = time.clock()
done_looping = False
epoch = 0
while(epoch < training_epochs) and (not done_looping):
epoch += 1
for minibatch_index in range(n_train_batches):
minibatch_avg_cost = train_fn(minibatch_index)
iter - (epoch - 1) * n_train_batches + minibatch_index
if (iter+1) % validation_frequency == 0:
validation_losses = valid_model()
this_validation_loss = numpy.mean(validation_losses)
print('epoch %i, minibatch %i/%i, validation error %f %%' %
(epoch, minibatch_index+1, n_train_batches,
this_validation_loss*100.)
)
if(this_validation_loss < best_validation_loss):
if(this_validation_loss < best_validation*improvement_threshold):
patience = max(patience, iter*patience_increase)
# save best validation score and iteration number
best_validation_loss = this_validation_loss
best_iter = iter
# test it on the test set
test_score = test_model()
test_score = numpy.mean(test_score)
print(' epoch %i, minibatch %i/%i, test error of '
'best model %f %%' %
(epoch, minibatch_index,n_train_batches,
test_score*100.))
if patience <= iter:
done_looping = True
break
end_time = time.clock()
print('Optimization complete with best validation score of %f %%, '
'on iteration %i, '
'with test performance %f %%'%
(best_validation_loss,best_iter+1,test_score*100.)
)
print('The training code for file ' +
os.path.split(__file__)[1] +
' ran for #.2fs'%(end_time - start_time)
)
def test_dA(learning_rate=0.1, training_epochs=75,
dataset='mnist.pkl.gz',
batch_size=20, output_folder='dA_plots'):
"""
This demo is tested on MNIST
:type learning_rate: float
:param learning_rate: learning rate used for training the DeNosing
AutoEncoder
:type training_epochs: int
:param training_epochs: number of epochs used for training
:type dataset: string
:param dataset: path to the picked dataset
"""
datasets = load_data(dataset)
train_set_x, train_set_y = datasets[0]
# compute number of minibatches for training, validation and testing
n_train_batches = train_set_x.get_value(borrow=True).shape[0] / batch_size
# start-snippet-2
# allocate symbolic variables for the data
index = T.lscalar() # index to a [mini]batch
x = T.matrix('x') # the data is presented as rasterized images
# end-snippet-2
if not os.path.isdir(output_folder):
os.makedirs(output_folder)
os.chdir(output_folder)
####################################
# BUILDING THE MODEL NO CORRUPTION #
####################################
rng = numpy.random.RandomState(123)
theano_rng = RandomStreams(rng.randint(2 ** 30))
da = dA(
numpy_rng=rng,
theano_rng=theano_rng,
input=x,
n_visible=28 * 28,
n_hidden=500
)
cost, updates = da.get_cost_updates(
corruption_level=0.,
learning_rate=learning_rate
)
train_da = theano.function(
[index],
cost,
updates=updates,
givens={
x: train_set_x[index * batch_size: (index + 1) * batch_size]
}
)
start_time = timeit.default_timer()
############
# TRAINING #
############
# go through training epochs
for epoch in xrange(training_epochs):
# go through trainng set
c = []
for batch_index in xrange(n_train_batches):
c.append(train_da(batch_index))
print 'Training epoch %d, cost ' % epoch, numpy.mean(c)
end_time = timeit.default_timer()
training_time = (end_time - start_time)
print >> sys.stderr, ('The no corruption code for file ' +
os.path.split(__file__)[1] +
' ran for %.2fm' % ((training_time) / 60.))
image = Image.fromarray(
tile_raster_images(X=da.W.get_value(borrow=True).T,
img_shape=(28, 28), tile_shape=(10, 10),
tile_spacing=(1, 1)))
image.save('filters_corruption_0.png')
# start-snippet-3
#####################################
# BUILDING THE MODEL CORRUPTION 30% #
#####################################
rng = numpy.random.RandomState(123)
theano_rng = RandomStreams(rng.randint(2 ** 30))
da = dA(
numpy_rng=rng,
theano_rng=theano_rng,
input=x,
n_visible=28 * 28,
n_hidden=500
)
cost, updates = da.get_cost_updates(
corruption_level=0.6,
learning_rate=learning_rate
)
train_da = theano.function(
[index],
cost,
updates=updates,
givens={
x: train_set_x[index * batch_size: (index + 1) * batch_size]
}
)
start_time = timeit.default_timer()
############
# TRAINING #
############
# go through training epochs
for epoch in xrange(training_epochs):
# go through trainng set
c = []
for batch_index in xrange(n_train_batches):
c.append(train_da(batch_index))
print 'Training epoch %d, cost ' % epoch, numpy.mean(c)
end_time = timeit.default_timer()
training_time = (end_time - start_time)
print >> sys.stderr, ('The 30% corruption code for file ' +
os.path.split(__file__)[1] +
' ran for %.2fm' % (training_time / 60.))
# end-snippet-3
# start-snippet-4
image = Image.fromarray(tile_raster_images(
X=da.W.get_value(borrow=True).T,
img_shape=(28, 28), tile_shape=(10, 10),
tile_spacing=(1, 1)))
image.save('filters_corruption_30.png')
# end-snippet-4
os.chdir('../')
def test_DBN(finetune_lr=0.1,
pretraining_epochs=100,
pretrain_lr=0.01,
k=1,
training_epochs=1000,
dataset='d:/mnist.pkl.gz',
batch_size=500):
"""
Demonstrates how to train and test a Deep Belief Network.
This is demonstrated on MNIST.
:type learning_rate: float
:param learning_rate: learning rate used in the finetune stage
:type pretraining_epochs: int
:param pretraining_epochs: number of epoch to do pretraining
:type pretrain_lr: float
:param pretrain_lr: learning rate to be used during pre-training
:type k: int
:param k: number of Gibbs steps in CD/PCD
:type training_epochs: int
:param training_epochs: maximal number of iterations ot run the optimizer
:type dataset: string
:param dataset: path the the pickled dataset
:type batch_size: int
:param batch_size: the size of a minibatch
"""
theano.config.blas.ldflags = "-lopenblas"
datasets = load_data(dataset)
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_train_batches = train_set_x.get_value(borrow=True).shape[0] / batch_size
# numpy random generator
numpy_rng = numpy.random.RandomState(123)
print '... building the model'
# construct the Deep Belief Network
dbn = DBN(numpy_rng=numpy_rng,
n_ins=28 * 28,
hidden_layers_sizes=[1000, 1000, 1000],
n_outs=10)
#########################
# PRETRAINING THE MODEL #
#########################
print '... getting the pretraining functions'
pretraining_fns = dbn.pretraining_functions(train_set_x=train_set_x,
batch_size=batch_size,
k=k)
print '... pre-training the model'
start_time = time.clock()
## Pre-train layer-wise
for i in xrange(dbn.n_layers):
# go through pretraining epochs
for epoch in xrange(pretraining_epochs):
# go through the training set
c = []
for batch_index in xrange(n_train_batches):
c.append(pretraining_fns[i](index=batch_index, lr=pretrain_lr))
print 'Pre-training layer %i, epoch %d, cost ' % (i, epoch),
print numpy.mean(c)
end_time = time.clock()
print >> sys.stderr, ('The pretraining code for file ' +
os.path.split(__file__)[1] + ' ran for %.2fm' %
((end_time - start_time) / 60.))
########################
# FINETUNING THE MODEL #
########################
# get the training, validation and testing function for the model
print '... getting the finetuning functions'
train_fn, validate_model, test_model = dbn.build_finetune_functions(
datasets=datasets, batch_size=batch_size, learning_rate=finetune_lr)
print '... finetunning the model'
# early-stopping parameters
patience = 4 * n_train_batches # look as this many examples regardless
patience_increase = 2. # wait this much longer when a new best is
# found
improvement_threshold = 0.995 # a relative improvement of this much is
# considered significant
validation_frequency = min(n_train_batches, patience / 2)
# go through this many
# minibatche before checking the network
# on the validation set; in this case we
# check every epoch
best_params = None
best_validation_loss = numpy.inf
test_score = 0.
start_time = time.clock()
done_looping = False
epoch = 0
while (epoch < training_epochs) and (not done_looping):
epoch = epoch + 1
for minibatch_index in xrange(n_train_batches):
minibatch_avg_cost = train_fn(minibatch_index)
iter = (epoch - 1) * n_train_batches + minibatch_index
if (iter + 1) % validation_frequency == 0:
validation_losses = validate_model()
this_validation_loss = numpy.mean(validation_losses)
print('epoch %i, minibatch %i/%i, validation error %f %%' % \
(epoch, minibatch_index + 1, n_train_batches,
this_validation_loss * 100.))
# if we got the best validation score until now
if this_validation_loss < best_validation_loss:
#improve patience if loss improvement is good enough
if (this_validation_loss <
best_validation_loss * improvement_threshold):
patience = max(patience, iter * patience_increase)
# save best validation score and iteration number
best_validation_loss = this_validation_loss
best_iter = iter
# test it on the test set
test_losses = test_model()
test_score = numpy.mean(test_losses)
print((' epoch %i, minibatch %i/%i, test error of '
'best model %f %%') %
(epoch, minibatch_index + 1, n_train_batches,
test_score * 100.))
if patience <= iter:
done_looping = True
break
end_time = time.clock()
print(('Optimization complete with best validation score of %f %%,'
'with test performance %f %%') %
(best_validation_loss * 100., test_score * 100.))
print >> sys.stderr, ('The fine tuning code for file ' +
os.path.split(__file__)[1] + ' ran for %.2fm' %
((end_time - start_time) / 60.))
def testdA(learning_rate=0.1, training_epochs=15, dataset="mnist.pkl.gz", batch_size=20, output_folder="dA_plots"):
print("Loading Data" + "." * 20)
datasets = load_data(dataset)
trainSetX, trainSetY = datasets[0]
# validSetX, validSetY = datasets[1]
# trainSetX, testSetY = datasets[2]
n_train_batches = trainSetX.get_value(borrow=True).shape[0] // batch_size
index = T.lscalar()
x = T.matrix("x")
if not os.path.isdir(output_folder):
os.makedirs(output_folder)
os.chdir(output_folder)
print("Building the Model With No Corruption " + "." * 20)
rng = numpy.random.RandomState(123)
theano_rng = RandomStreams(rng.randint(2 ** 30))
da = dA(numpy_rng=rng, theano_rng=theano_rng, input=x, n_visible=28 * 28, n_hidden=500)
cost, updates = da.get_cost_updates(corruption_level=0.0, learning_rate=learning_rate)
train_da = theano.function(
inputs=[index],
outputs=cost,
updates=updates,
givens={x: trainSetX[index * batch_size : (index + 1) * batch_size]},
)
start_time = time.clock()
print("Training" + "." * 20)
for epoch in range(n_train_batches):
c = []
for batch_index in range(n_train_batches):
c.append(train_da(batch_index))
print("Training epoch %d, cost " % epoch, numpy.mean(c))
end_time = time.clock()
training_time = end_time - start_time
print("The no corruption code for file " + os.path.split(__file__)[1] + " ran for %.2fs" % training_time)
img = Image.fromarray(
tile_raster_images(
X=da.W.get_value(borrow=True).T, img_shape=(28, 28), tile_shape=(10, 10), tile_spacing=(1, 1)
)
)
img.save("filter_corrupting_0.png")
print("Building the Model With Corruption 30%" + "." * 20)
theano_rng = RandomStreams(rng.randint(2 ** 30))
rng = numpy.random.RandomState(123)
da = dA(numpy_rng=rng, theano_rng=theano_rng, input=x, n_visible=28 * 28, n_hidden=500)
cost, updates = da.get_cost_updates(corruption_level=0.3, learning_rate=learning_rate)
train_da = theano.function(
inputs=[index],
outputs=cost,
updates=updates,
givens={x: trainSetX[index * batch_size : (index + 1) * batch_size]},
)
start_time = time.clock()
print("Training" + "." * 20)
for epoch in range(n_train_batches):
c = []
for batch_index in range(n_train_batches):
c.append(train_da(batch_index))
print("Training epoch %d, cost " % epoch, numpy.mean(c))
end_time = time.clock()
training_time = end_time - start_time
print("The 30% corruption code for file " + os.path.split(__file__)[1] + " ran for %.2fs" % training_time)
img = Image.fromarray(
tile_raster_images(
X=da.W.get_value(borrow=True).T, img_shape=(28, 28), tile_shape=(10, 10), tile_spacing=(1, 1)
)
)
img.save("filter_corrupting_30.png")
os.chdir("../")