I have a question about the use of the sample_weight parameter in the context of data augmentation in Keras with the ImageDataGenerator. Let's say I have a series of simple images with just one class of objects. So, for each image, I will have a corresponding mask with pixels = 0 for the background and 1 for where the object is labeled.
However, this dataset is unbalanced because a significant amount of these images are empty, which mean with masks just containing 0.
If I understood well, the 'sample_weight' parameter of the flow method of ImageDataGenerator is here to put the focus on the the samples of my dataset that I find more interesting, i.e. where my object is present.
My question is: what is the concrete influence of this sample_weight parameter on the training of my model. Does it influence the data augmentation? If I use the 'validation_split' parameter, does it influence the way validation sets are generated?
Here is the part of my code my question refers to:
data_gen_args = dict(rotation_range=90,
width_shift_range=0.4,
height_shift_range=0.4,
zoom_range=0.4,
horizontal_flip=True,
fill_mode='reflect',
rescale=1. / 255,
validation_split=0.2,
data_format='channels_last'
)
image_datagen = ImageDataGenerator(**data_gen_args)
imf = image_datagen.flow(
x=stacked_images_channel,
y=stacked_masks_channel,
batch_size=batch_size,
shuffle=False,
seed=seed,subset='training',
sample_weight = sample_weight,
save_to_dir = 'traindir',
save_prefix = 'train_'
)
valf = image_datagen.flow(
x=stacked_images_channel,
y=stacked_masks_channel,
batch_size=batch_size,
shuffle=False,
seed=seed,subset='validation',
sample_weight = sample_weight,
save_to_dir = 'valdir',
save_prefix = 'val_'
)
STEP_SIZE_TRAIN=imf.n//imf.batch_size
STEP_SIZE_VALID=valf.n//valf.batch_size
model = unet.UNet2(numberOfClasses, imshape, '', learningRate, depth=4)
history = model.fit_generator(generator=imf,
steps_per_epoch=STEP_SIZE_TRAIN,
epochs=epochs,
validation_data=valf,
validation_steps=STEP_SIZE_VALID,
verbose=2
)
Thank you in advance for your attention.
As for Keras 2.2.5 with preprocessing at 1.1.0, the sample_weight is passed along with the samples and applied during processing. When calling .fit_generator, the model is trained on batches, each batch using sample weights:
model.train_on_batch(x, y,
sample_weight=sample_weight,
class_weight=class_weight)
In the source code of .train_on_batch, the documentation states: "sample_weight: Optional array of the same length as x, containing weights to apply to the model's loss for each sample. (...)". The actual application of weights happens when calculating loss on each batch. When compiling a model, Keras generates a "weighted loss" function out of the desired loss function. The weighted computation is stated in the code as:
def weighted(y_true, y_pred, weights, mask=None):
"""Wrapper function.
# Arguments
y_true: `y_true` argument of `fn`.
y_pred: `y_pred` argument of `fn`.
weights: Weights tensor.
mask: Mask tensor.
# Returns
Scalar tensor.
"""
# score_array has ndim >= 2
score_array = fn(y_true, y_pred)
if mask is not None:
# Cast the mask to floatX to avoid float64 upcasting in Theano
mask = K.cast(mask, K.floatx())
# mask should have the same shape as score_array
score_array *= mask
# the loss per batch should be proportional
# to the number of unmasked samples.
score_array /= K.mean(mask) + K.epsilon()
# apply sample weighting
if weights is not None:
# reduce score_array to same ndim as weight array
ndim = K.ndim(score_array)
weight_ndim = K.ndim(weights)
score_array = K.mean(score_array,
axis=list(range(weight_ndim, ndim)))
score_array *= weights
score_array /= K.mean(K.cast(K.not_equal(weights, 0), K.floatx()))
return K.mean(score_array)
This wrapper shows it first calculates the desired loss (call to fn(y_true, y_pred)), then applies weighing if weights where passed (either with sample_weight or class_weight).
With this context in mind:
what is the concrete influence of this sample_weight parameter on the training of my model.
Weights are basically multiplied to the loss (and normalized). So "heavy" weights (more than 1) samples cause more loss, so larger gradients. "Light" weights reduce the importance of the sample and lead to smaller gradients.
Does it influence the data augmentation?
It depends on what you mean. Here is what I can say from experience, where I perform augmentation before feeding a Keras data generator (doing so as there were issues in preprocessing, as far as I know still existing in Preprocessing 1.1.0):
When feeding already augmented data to the generator, the .flow call will require a sample weights list as long as the input data. So the influence of weighing on augmentation depends on how the weights are chosen. A data point augmented N times may assign the same weight to each augmentation, or 1/N depending on the intent.
The default behaviour in Keras seems to assign the same weight to each augmentation (transform) performed by Keras. The code looks pretty clear, although I have never relied on it.
If I use the 'validation_split' parameter, does it influence the way validation sets are generated?
The sample_weight parameter does not seem to interfere with validation_split. I have not looked into the code specifically, but splitting basically gets the input data, and keeps a split for validation---whatever the data is. When sample_weight is added, what changes is each data point: Without weight, data is (x, y); with weight, data becomes (x, y, weight).
Related
I am getting acquainted with Tensorflow-Probability and here I am running into a problem. During training, the model returns nan as the loss (possibly meaning a huge loss that causes overflowing). Since the functional form of the synthetic data is not overly complicated and the ratio of data points to parameters is not frightening at first glance at least I wonder what is the problem and how it could be corrected.
The code is the following --accompanied by some possibly helpful images:
# Create and plot 5000 data points
x_train = np.linspace(-1, 2, 5000)[:, np.newaxis]
y_train = np.power(x_train, 3) + 0.1*(2+x_train)*np.random.randn(5000)[:, np.newaxis]
plt.scatter(x_train, y_train, alpha=0.1)
plt.show()
# Define the prior weight distribution -- all N(0, 1) -- and not trainable
def prior(kernel_size, bias_size, dtype = None):
n = kernel_size + bias_size
prior_model = Sequential([
tfpl.DistributionLambda(
lambda t: tfd.MultivariateNormalDiag(loc = tf.zeros(n) , scale_diag = tf.ones(n)
))
])
return(prior_model)
# Define variational posterior weight distribution -- multivariate Gaussian
def posterior(kernel_size, bias_size, dtype = None):
n = kernel_size + bias_size
posterior_model = Sequential([
tfpl.VariableLayer(tfpl.MultivariateNormalTriL.params_size(n) , dtype = dtype), # The parameters of the model are declared Variables that are trainable
tfpl.MultivariateNormalTriL(n) # The posterior function will return to the Variational layer that will call it a MultivariateNormalTril object that will have as many dimensions
# as the parameters of the Variational Dense Layer. That means that each parameter will be generated by a distinct Normal Gaussian shifted and scaled
# by a mu and sigma learned from the data, independently of all the other weights. The output of this Variablelayer will become the input to the
# MultivariateNormalTriL object.
# The shape of the VariableLayer object will be defined by the number of paramaters needed to create the MultivariateNormalTriL object given
# that it will live in a Space of n dimensions (event_size = n). This number is returned by the tfpl.MultivariateNormalTriL.params_size(n)
])
return(posterior_model)
x_in = Input(shape = (1,))
x = tfpl.DenseVariational(units= 2**4,
make_prior_fn=prior,
make_posterior_fn=posterior,
kl_weight=1/x_train.shape[0],
activation='relu')(x_in)
x = tfpl.DenseVariational(units= 2**4,
make_prior_fn=prior,
make_posterior_fn=posterior,
kl_weight=1/x_train.shape[0],
activation='relu')(x)
x = tfpl.DenseVariational(units=tfpl.IndependentNormal.params_size(1),
make_prior_fn=prior,
make_posterior_fn=posterior,
kl_weight=1/x_train.shape[0])(x)
y_out = tfpl.IndependentNormal(1)(x)
model = Model(inputs = x_in, outputs = y_out)
def nll(y_true, y_pred):
return -y_pred.log_prob(y_true)
model.compile(loss=nll, optimizer= 'Adam')
model.summary()
Train the model
history = model.fit(x_train1, y_train1, epochs=500)
The problem seems to be in the loss function: negative log-likelihood of the independent normal distribution without any specified location and scale leads to the untamed variance which leads to the blowing up the final loss value. Since you're experimenting with the variational layers, you must be interested in the estimation of the epistemic uncertainty, to that end, I'd recommend to apply the constant variance.
I tried to make a couple of slight changes to your code within the following lines:
first of all, the final output y_out comes directly from the final variational layer without any IndpendnetNormal distribution layer:
y_out = tfpl.DenseVariational(units=1,
make_prior_fn=prior,
make_posterior_fn=posterior,
kl_weight=1/x_train.shape[0])(x)
second, the loss function now contains the necessary calculations with the normal distribution you need but with the static variance in order to avoid the blowing up of the loss during training:
def nll(y_true, y_pred):
dist = tfp.distributions.Normal(loc=y_pred, scale=1.0)
return tf.reduce_sum(-dist.log_prob(y_true))
then the model is compiled and trained in the same way as before:
model.compile(loss=nll, optimizer= 'Adam')
history = model.fit(x_train, y_train, epochs=3000)
and finally let's sample 100 different predictions from the trained model and plot these values to visualize the epistemic uncertainty of the model:
predicted = [model(x_train) for _ in range(100)]
for i, res in enumerate(predicted):
plt.plot(x_train, res , alpha=0.1)
plt.scatter(x_train, y_train, alpha=0.1)
plt.show()
After 3000 epochs the result looks like this (with the reduced number of training points to 3000 instead of 5000 to speed-up the training):
The model has 38,589 trainable parameters but you have only 5,000 points as data; so, effective training is impossible with so many parameters.
According to Keras Code, Keras computes the loss value considering optional weights and masks.
for i in range(len(self.outputs)):
if i in skip_target_indices:
continue
y_true = self.targets[i]
y_pred = self.outputs[i]
weighted_loss = weighted_losses[i]
sample_weight = sample_weights[i]
mask = masks[i]
loss_weight = loss_weights_list[i]
with K.name_scope(self.output_names[i] + '_loss'):
output_loss = weighted_loss(y_true, y_pred,
sample_weight, mask)
if len(self.outputs) > 1:
self.metrics_tensors.append(output_loss)
self.metrics_names.append(self.output_names[i] + '_loss')
if total_loss is None:
total_loss = loss_weight * output_loss
else:
total_loss += loss_weight * output_loss
On the other hand, in Keras documentation I see the basic loss function is introduced in compile function, and then sample or class weights can be introduced in fit command.
I am not sure how to relate 'weights and masks' in the first code to 'sample and class weights' in the second document. Can anybody give me more explanation?
My application is actually a Convolutional LSTM network, where I input a series of images and want the network to produce an output map (with the same size of input maps) of pixel classes, but some pixels don't have valid labels during training. Should I use weight or mask, sample or class?
I have to train a network on unlabelled data of binary type (True/False), which sounds like unsupervised learning. This is what the normalised data look like:
array([[-0.05744527, -1.03575495, -0.1940105 , -1.15348956, -0.62664491,
-0.98484037],
[-0.05497629, -0.50935675, -0.19396862, -0.68990988, -0.10551919,
-0.72375012],
[-0.03275552, 0.31480204, -0.1834951 , 0.23724946, 0.15504367,
0.29810553],
...,
[-0.05744527, -0.68482282, -0.1940105 , -0.87534175, -0.23580062,
-0.98484037],
[-0.05744527, -1.50366446, -0.1940105 , -1.52435329, -1.14777063,
-0.98484037],
[-0.05744527, -1.26970971, -0.1940105 , -1.33892142, -0.88720777,
-0.98484037]])
However, I do have a constraint on the total number of True labels in my data. This doesn't mean I can build a classical custom loss function in Keras taking (y_true, y_pred) arguments as required: my external constraint is just on the predicted total of True and False, not on the individual labels.
My question is whether there is a somewhat "standard" approach to this kind of problems, and how that is implementable in Keras.
POSSIBLE SOLUTION
Should I assign y_true randomly as 0/1, have a network return y_pred as 1/0 with a sigmoid activation function, and then define my loss function as
sum_y_true = 500 # arbitrary constant known a priori
def loss_function(y_true, y_pred):
loss = np.abs(y_pred.sum() - sum_y_true)
return loss
In the end, I went with the following solution, which worked.
1) Define batches in your dataframe df with a batch_id column, so that in each batch Y_train is your identical "batch ground truth" (in my case, the total number of True labels in the batch). You can then pass these instances together to the network. This can be done with a generator:
def grouper(g,x,y):
while True:
for gr in g.unique():
# this assigns indices to the entire set of values in g,
# then subsects to all the rows in which g == gr
indices = g == gr
yield (x[indices],y[indices])
# train set
train_generator = grouper(df.loc[df['set'] == 'train','batch_id'], X_train, Y_train)
# validation set
val_generator = grouper(df.loc[df['set'] == 'val','batch_id'], X_val, Y_val)
2) define a custom loss function, to track how close the total number of instances predicted as true matches the ground truth:
def custom_delta(y_true, y_pred):
loss = K.abs(K.mean(y_true) - K.sum(y_pred))
return loss
def custom_wrapper():
def custom_loss_function(y_true, y_pred):
return custom_delta(y_true, y_pred)
return custom_loss_function
Note that here
a) Each y_true label is already the sum of the ground truth in our batch (cause we don't have individual values). That's why y_true is not summed over;
b) K.mean is actually a bit of an overkill to extract a single scalar from this uniform tensor, in which all y_true values in each batch are identical - K.min or K.max would also work, but I haven't tested whether their performance is faster.
3) Use fit_generator instead of fit:
fmodel = Sequential()
# ...your layers...
# Create the loss function object using the wrapper function above
loss_ = custom_wrapper()
fmodel.compile(loss=loss_, optimizer='adam')
history1 = fmodel.fit_generator(train_generator, steps_per_epoch=total_batches,
validation_data=val_generator,
validation_steps=df.loc[encs.df['set'] == 'val','batch_id'].nunique(),
epochs=20, verbose = 2)
This way the problem is basically addressed as one of supervised learning, although without individual labels, which means that notions like true/false positive are meaningless here.
This approach not only managed to give me a y_pred that closely matches the totals I know per batch. It actually finds two groups (True/False) that occupy the expected different portions of parameter space.
Can someone tell me mathematically how sample_weight and class_weight are used in Keras in the calculation of loss function and metrics? A simple mathematical express will be great.
It is a simple multiplication. The loss contributed by the sample is magnified by its sample weight. Assuming i = 1 to n samples, a weight vector of sample weights w of length n, and that the loss for sample i is denoted L_i:
In Keras in particular, the product of each sample's loss with its weight is divided by the fraction of weights that are not 0 such that the loss per batch is proportional to the number of weight > 0 samples. Let p be the proportion of non-zero weights.
Here's the relevant snippet of code from the Keras repo:
score_array = loss_fn(y_true, y_pred)
if weights is not None:
score_array *= weights
score_array /= K.mean(K.cast(K.not_equal(weights, 0), K.floatx()))
return K.mean(score_array)
class_weight is used in the same way as sample_weight; it is just provided as a convenience to specify certain weights across entire classes.
The sample weights are currently not applied to metrics, only loss.
Keras documentation isn't clear what this actually is. I understand we can use this to compress the input feature space into a smaller one. But how is this done from a neural design perspective? Is it an autoenocder, RBM?
As far as I know, the Embedding layer is a simple matrix multiplication that transforms words into their corresponding word embeddings.
The weights of the Embedding layer are of the shape (vocabulary_size, embedding_dimension). For each training sample, its input are integers, which represent certain words. The integers are in the range of the vocabulary size. The Embedding layer transforms each integer i into the ith line of the embedding weights matrix.
In order to quickly do this as a matrix multiplication, the input integers are not stored as a list of integers but as a one-hot matrix. Therefore the input shape is (nb_words, vocabulary_size) with one non-zero value per line. If you multiply this by the embedding weights, you get the output in the shape
(nb_words, vocab_size) x (vocab_size, embedding_dim) = (nb_words, embedding_dim)
So with a simple matrix multiplication you transform all the words in a sample into the corresponding word embeddings.
The Keras Embedding layer is not performing any matrix multiplication but it only:
1. creates a weight matrix of (vocabulary_size)x(embedding_dimension) dimensions
2. indexes this weight matrix
It is always useful to have a look at the source code to understand what a class does. In this case, we will have a look at the class Embedding which inherits from the base layer class called Layer.
(1) - Creating a weight matrix of (vocabulary_size)x(embedding_dimension) dimensions:
This is occuring at the build function of Embedding:
def build(self, input_shape):
self.embeddings = self.add_weight(
shape=(self.input_dim, self.output_dim),
initializer=self.embeddings_initializer,
name='embeddings',
regularizer=self.embeddings_regularizer,
constraint=self.embeddings_constraint,
dtype=self.dtype)
self.built = True
If you have a look at the base class Layer you will see that the function add_weight above simply creates a matrix of trainable weights (in this case of (vocabulary_size)x(embedding_dimension) dimensions):
def add_weight(self,
name,
shape,
dtype=None,
initializer=None,
regularizer=None,
trainable=True,
constraint=None):
"""Adds a weight variable to the layer.
# Arguments
name: String, the name for the weight variable.
shape: The shape tuple of the weight.
dtype: The dtype of the weight.
initializer: An Initializer instance (callable).
regularizer: An optional Regularizer instance.
trainable: A boolean, whether the weight should
be trained via backprop or not (assuming
that the layer itself is also trainable).
constraint: An optional Constraint instance.
# Returns
The created weight variable.
"""
initializer = initializers.get(initializer)
if dtype is None:
dtype = K.floatx()
weight = K.variable(initializer(shape),
dtype=dtype,
name=name,
constraint=constraint)
if regularizer is not None:
with K.name_scope('weight_regularizer'):
self.add_loss(regularizer(weight))
if trainable:
self._trainable_weights.append(weight)
else:
self._non_trainable_weights.append(weight)
return weight
(2) - Indexing this weight matrix
This is occuring at the call function of Embedding:
def call(self, inputs):
if K.dtype(inputs) != 'int32':
inputs = K.cast(inputs, 'int32')
out = K.gather(self.embeddings, inputs)
return out
This functions returns the output of the Embedding layer which is K.gather(self.embeddings, inputs). What tf.keras.backend.gather exactly does is to index the weights matrix self.embeddings (see build function above) according to the inputs which should be lists of positive integers.
These lists can be retrieved for example if you pass your text/words inputs to the one_hot function of Keras which encodes a text into a list of word indexes of size n (this is NOT one hot encoding - see also this example for more info: https://machinelearningmastery.com/use-word-embedding-layers-deep-learning-keras/).
Therefore, that's all. There is no matrix multiplication.
On the contrary, the Keras Embedding layer is only useful because exactly it avoids performing a matrix multiplication and hence it economizes on some computational resources.
Otherwise, you could just use a Keras Dense layer (after you have encoded your input data) to get a matrix of trainable weights (of (vocabulary_size)x(embedding_dimension) dimensions) and then simply do the multiplication to get the output which will be exactly the same with the output of the Embedding layer.
In Keras, the Embedding layer is NOT a simple matrix multiplication layer, but a look-up table layer (see call function below or the original definition).
def call(self, inputs):
if K.dtype(inputs) != 'int32':
inputs = K.cast(inputs, 'int32')
out = K.gather(self.embeddings, inputs)
return out
What it does is to map each a known integer n in inputs to a trainable feature vector W[n], whose dimension is the so-called embedded feature length.
In simple words (from the functionality point of view), it is a one-hot encoder and fully-connected layer. The layer weights are trainable.