Using Keras from Tensorflow 1.4.1, how does one copy weights from one model to another?
As some background, I'm trying to implement a deep-q network (DQN) for Atari games following the DQN publication by DeepMind. My understanding is that the implementation uses two networks, Q and Q'. The weights of Q are trained using gradient descent, and then the weights are copied periodically to Q'.
Here's how I build Q and Q':
ACT_SIZE = 4
LEARN_RATE = 0.0025
OBS_SIZE = 128
def buildModel():
model = tf.keras.models.Sequential()
model.add(tf.keras.layers.Lambda(lambda x: x / 255.0, input_shape=OBS_SIZE))
model.add(tf.keras.layers.Dense(128, activation="relu"))
model.add(tf.keras.layers.Dense(128, activation="relu"))
model.add(tf.keras.layers.Dense(ACT_SIZE, activation="linear"))
opt = tf.keras.optimizers.RMSprop(lr=LEARN_RATE)
model.compile(loss="mean_squared_error", optimizer=opt)
return model
I call that twice to get Q and Q'.
I have an updateTargetModel method below that is my attempt at copying weights. The code runs fine, but my overall DQN implementation is failing. I'm really just trying to verify if this is a valid way of copying weights from one network to another.
def updateTargetModel(model, targetModel):
modelWeights = model.trainable_weights
targetModelWeights = targetModel.trainable_weights
for i in range(len(targetModelWeights)):
targetModelWeights[i].assign(modelWeights[i])
There's another question here that discusses saving and loading weights to and from disk (Tensorflow Copy Weights Issue), but there's no accepted answer. There is also a question about loading weights from individual layers (Copying weights from one Conv2D layer to another), but I'm wanting to copy the entire model's weights.
Actually what you've done is much more than simply copying weights. You made these two models identical all the time. Every time you update one model - the second one is also updated - as both models have the same weights variables.
If you want to just copy weights - the simplest way is by this command:
target_model.set_weights(model.get_weights())
Related
I am trying to train a classifier based on the InceptionV3 architecture in Keras.
For this I loaded the pre-trained InceptionV3 model, without top, and added a final fully connected layer for the classes of my classification problem. In the first training I froze the InceptionV3 base model and only trained the final fully connected layer.
In the second step I want to "fine tune" the network by unfreezing a part of the InceptionV3 model.
Now I know that the InceptionV3 model makes extensive use of BatchNorm layers. It is recommended (link to documentation), when BatchNorm layers are "unfrozen" for fine tuning when transfer learning, to keep the mean and variances as computed by the BatchNorm layers fixed. This should be done by setting the BatchNorm layers to inference mode instead of training mode.
Please also see: What's the difference between the training argument in call() and the trainable attribute?
Now my main question is: how to set ONLY the BatchNorm layers of the InceptionV3 model to inference mode?
Currently I set the whole InceptionV3 base model to inference mode by setting the "training" argument when assembling the network:
inputs = keras.Input(shape=input_shape)
# Scale the 0-255 RGB values to 0.0-1.0 RGB values
x = layers.experimental.preprocessing.Rescaling(1./255)(inputs)
# Set include_top to False so that the final fully connected (with pre-loaded weights) layer is not included.
# We will add our own fully connected layer for our own set of classes to the network.
base_model = keras.applications.InceptionV3(input_shape=input_shape, weights='imagenet', include_top=False)
x = base_model(x, training=False)
# Classification block
x = layers.GlobalAveragePooling2D(name='avg_pool')(x)
x = layers.Dense(num_classes, activation='softmax', name='predictions')(x)
model = keras.Model(inputs=inputs, outputs=x)
What I don't like about this, is that in this way I set the whole model to inference mode which may set some layers to inference mode which should not be.
Here is the part of the code that loads the weights from the initial training that I did and the code that freezes the first 150 layers and unfreezes the remaining layers of the InceptionV3 part:
model.load_weights(load_model_weight_file_name)
for layer in base_model.layers[: 150]:
layer.trainable = False
for layer in base_model.layers[ 150:]:
layer.trainable = True
The rest of my code (not shown here) are the usual compile and fit calls.
Running this code seems to result a network that doesn't really learn (loss and accuracy remain approximately the same). I tried different orders of magnitude for the optimization step size, but that doesn't seem to help.
Another thing that I observed it that when I make the whole InceptionV3 part trainable
base_model.trainable = True
that the training starts with an accuracy server orders of magnitude smaller than were my first training round finished (and of course a much higher loss). Can someone explain this to me? I would at least expect the training to continue were it left off in terms of accuracy and loss.
You could do something like:
for layer in base_model.layers:
if isinstance(layer ,tf.keras.layers.BatchNormalization):
layer.trainable=False
This will iterate over each layer and check the type, setting to inference mode if the layer is BatchNorm.
As for the low starting accuracy during transfer learning, you're only loading the weights and not the optimiser state (as would occur with a full model.load() which loads architecture, weights, optimiser state etc).
This doesn't mean there's an error, but if you must load weights only just let it train, the optimiser will configure eventually and you should see progress. Also as you're potentially over-writing the pre-trained weights in your second run, make sure you use a lower learning rate so the updates are small in comparison i.e. fine-tune the weights rather than blast them to pieces.
How can I use the weights of a pre-trained network in my tensorflow project?
I know some theory information about this but no information about coding in tensorflow.
As been pointed out by #Matias Valdenegro in the comments, your first question does not make sense. For your second question however, there are multiple ways to do so. The term that you're searching for is Transfer Learning (TL). TL means transferring the "knowledge" (basically it's just the weights) from a pre-trained model into your model. Now there are several types of TL.
1) You transfer the entire weights from a pre-trained model into your model and use that as a starting point to train your network.
This is done in a situation where you now have extra data to train your model but you don't want to start over the training again. Therefore you just load the weights from your previous model and resume the training.
2) You transfer only some of the weights from a pre-trained model into your new model.
This is done in a situation where you have a model trained to classify between, say, 5 classes of objects. Now, you want to add/remove a class. You don't have to re-train the whole network from the start if the new class that you're adding has somewhat similar features with (an) existing class(es). Therefore, you build another model with the same exact architecture as your previous model except the fully-connected layers where now you have different output size. In this case, you'll want to load the weights of the convolutional layers from the previous model and freeze them while only re-train the fully-connected layers.
To perform these in Tensorflow,
1) The first type of TL can be performed by creating a model with the same exact architecture as the previous model and simply loading the model using tf.train.Saver().restore() module and continue the training.
2) The second type of TL can be performed by creating a model with the same exact architecture for the parts where you want to retain the weights and then specify the name of the weights in which you want to load from the previous pre-trained weights. You can use the parameter "trainable=False" to prevent Tensorflow from updating them.
I hope this helps.
I'm developing a machine learning model using keras and I notice that the available losses functions are not giving the best results on my test set.
I am using an Unet architecture, where I input a (16,16,3) image and the net also outputs a (16,16,3) picture (auto-encoder). I notice that maybe one way to improve the model would be if I used a loss function that compares pixel to pixel on the gradients (laplacian) between the net output and the ground truth. However, I did not found any tutorial that would handle this kind of application, because it would need to use opencv laplacian function on each output image from the net.
The loss function would be something like this:
def laplacian_loss(y_true, y_pred):
# y_true already is the calculated gradients, only needs to compute on the y_pred
# calculates the gradients for each predicted image
y_pred_lap = []
for img in y_pred:
laplacian = cv2.Laplacian( np.float64(img), cv2.CV_64F )
y_pred_lap.append( laplacian )
y_pred_lap = np.array(y_pred_lap)
# mean squared error, according to keras losses documentation
return K.mean(K.square(y_pred_lap - y_true), axis=-1)
Has anyone done something like that for loss calculation?
Given the code above, it seems that it would be equivalent to using a Lambda() layer as the output layer that applies that transformation in the image, before considering the mean square error.
Regardless as whether it is implemented as a Lambda() layer or in the loss function; the transformation needs to be such that Tensorflow understands how to calculate the gradients. The simplest was to do this would probably be to reimplement the cv2.Laplacian computation using Tensorflow math operations.
In order to use the cv2 library directly, you need to create a function that calculates the gradients for what happens inside the cv2 lib; that seems significantly more error prone.
Gradient descent optimisation relies on being able to compute gradients from the inputs to the loss; and back. Any operation in the middle must be differentiable; and Tensorflow must understand the math operations for auto differentiation to work; or you need to add them manually.
I managed to reach a easy solution. The main feature was that the gradient calculation is actually a 2D filter. For more information about it, please follow the link about the laplacian kernel. In that matter, is necessary that the output of my network be filtered by the laplacian kernel. For that, I created an extra convolutional layer with fixed weights, exactly as the laplacian kernel. After that, the network will have two outputs (one been the desired image, and the other been the gradient's image). So, is also necessary to define both losses.
To make it clearer, I'll exemplify. In the end of the network you'll have something like:
channels = 3 # number of channels of network output
lap = Conv2D(channels , (3,3), padding='same', name='laplacian') (net_output)
model = Model(inputs=[net_input], outputs=[net_out, lap])
Define how you want to calculate the losses for each output:
# losses for output, laplacian and gaussian
losses = {
"enhanced": "mse",
"laplacian": "mse"
}
lossWeights = {"enhanced": 1.0, "laplacian": 0.6}
Compile the model:
model.compile(optimizer=Adam(), loss=losses, loss_weights=lossWeights)
Define the laplacian kernel, apply its values in the weights of the above convolutional layer and set trainable equals False (so it won't be updated).
bias = np.asarray([0]*3)
# laplacian kernel
l = np.asarray([
[[[1,1,1],
[1,-8,1],
[1,1,1]
]]*channels
]*channels).astype(np.float32)
bias = np.asarray([0]*3).astype(np.float32)
wl = [l,bias]
model.get_layer('laplacian').set_weights(wl)
model.get_layer('laplacian').trainable = False
When training, remember that you need two values for the ground truth:
model.fit(x=X, y = {"out": y_out, "laplacian": y_lap})
Observation: Do not utilize the BatchNormalization layer! In case you use it, the weights in the laplacian layer will be updated!
I have 10 class dataset with this I got 85% accuracy, got the same accuracy on a saved model.
now I want to add a new class, how to add a new class To the saved model.
I tried by deleting the last layer and train but model get overfit and in prediction every Images show same result (newly added class).
This is what I did
model.pop()
base_model_layers = model.output
pred = Dense(11, activation='softmax')(base_model_layers)
model = Model(inputs=model.input, outputs=pred)
# compile and fit step
I have trained model with 10 class I want to load the model train with class 11 data and give predictions.
Using the model.pop() method and then the Keras Model() API will lead you to an error. The Model() API does not have the .pop() method, so if you want to re-train your model more than once you will have this error.
But the error only occurs if you, after the re-training, save the model and use the new saved model in the next re-training.
Another very wrong and used approach is to use the model.layers.pop(). This time the problem is that function only removes the last layer in the copy it returns. So, the model still has the layer, and just the method's return does not have the layer.
I recommend the following solution:
Admitting you have your already trained model saved in the model variable, something like:
model = load_my_trained_model_function()
# creating a new model
model_2 = Sequential()
# getting all the layers except the output one
for layer in model.layers[:-1]: # just exclude last layer from copying
model_2.add(layer)
# prevent the already trained layers from being trained again
# (you can use layers[:-n] to only freeze the model layers until the nth layer)
for layer in model_2.layers:
layer.trainable = False
# adding the new output layer, the name parameter is important
# otherwise, you will add a Dense_1 named layer, that normally already exists, leading to an error
model_2.add(Dense(num_neurons_you_want, name='new_Dense', activation='softmax'))
Now you should specify the compile and fit methods to train your model and it's done:
model_2.compile(loss='categorical_crossentropy',
optimizer='adam',
metrics=['accuracy'])
# model.fit trains the model
model_history = model_2.fit(x_train, y_train,
batch_size=batch_size,
epochs=epochs,
verbose=1,
validation_split=0.1)
EDIT:
Note that by adding a new output layer we do not have the weights and biases adjusted in the last training.
Thereby we lost pretty much everything from the previous training.
We need to save the weights and biases of the output layer of the previous training, and then we must add them to the new output layer.
We also must think if we should let all the layers train or not, or even if we should allow the training of only some intercalated layers.
To get the weights and biases from the output layer using Keras we can use the following method:
# weights_training[0] = layer weights
# weights_training[1] = layer biases
weights_training = model.layers[-1].get_weights()
Now you should specify the weights for the new output layer. You can use, for example, the mean of the weights for the weights of the new classes. It's up to you.
To set the weights and biases of the new output layer using Keras we can use the following method:
model_2.layers[-1].set_weights(weights_re_training)
model.pop()
base_model_layers = model.output
pred = Dense(11, activation='softmax')(base_model_layers)
model = Model(inputs=model.input, outputs=pred)
Freeze the first layers, before train it
for layer in model.layers[:-2]:
layer.trainable = False
I am assuming that the problem is singlelabel-multiclass classification i.e. a sample will belong to only 1 of the 11 classes.
This answer will be completely based on implementing the way humans learn into machines. Hence, this will not provide you with a proper code of how to do that but it will tell you what to do and you will be able to easily implement it in keras.
How does a human child learn when you teach him new things? At first, we ask him to forget the old and learn the new. This does not actually mean that the old learning is useless but it means that for the time while he is learning the new, the old knowledge should not interfere as it will confuse the brain. So, the child will only learn the new for sometime.
But the problem here is, things are related. Suppose, the child learned C programming language and then learned compilers. There is a relation between compilers and programming language. The child cannot master computer science if he learns these subjects separately, right? At this point we introduce the term 'intelligence'.
The kid who understands that there is a relation between the things he learned before and the things he learned now is 'intelligent'. And the kid who finds the actual relation between the two things is 'smart'. (Going deep into this is off-topic)
What I am trying to say is:
Make the model learn the new class separately.
And then, make the model find a relation between the previously learned classes and the new class.
To do this, you need to train two different models:
The model which learns to classify on the new class: this model will be a binary classifier. It predicts a 1 if the sample belongs to class 11 and 0 if it doesn't. Now, you already have the training data for samples belonging to class 11 but you might not have data for the samples which doesn't belong to class 11. For this, you can randomly select samples which belong to classes 1 to 10. But note that the ratio of samples belonging to class 11 to that not belonging to class 11 must be 1:1 in order to train the model properly. That means, 50% of the samples must belong to class 11.
Now, you have two separate models: the one which predicts class 1-10 and one which predicts class 11. Now, concatenate the outputs of (the 2nd last layers) these two models with a newly created Dense layer with 11 nodes and let the whole model retrain itself adjusting the weights of pretrained two models and learning new weights of the dense layer. Keep the learning rate low.
The final model is the third model which is a combination of two models (without last Dense layer) + a new Dense layer.
Thank you..
I haven't used Keras and I'm thinking whether to use it or not.
I want to save a trained layer to use later. For example:
I train a model.
Then, I gain a trained layer t_layer.
I have another model to train which consists of layer1, layer2, layer3 .
I want to use t_layer as layer2 and not to update this layer(i.e. t_layer does not learn any more).
This may be an odd attempt, but I want to try this. Is this possible on Keras?
Yes, it is.
You will probably have to save the layer's weights and biases instead of saving the layer itself, but it's possible.
Keras also allows you to save entire models.
Suppose you have a model in the var model:
weightsAndBiases = model.layers[i].get_weights()
This is a list of numpy arrays, very probably with two arrays: weighs and biases. You can simply use numpy.save() to save these two arrays and later you can create a similar layer and give it the weights:
from keras.layers import *
from keras.models import Model
inp = Input(....)
out1 = SomeKerasLayer(...)(inp)
out2 = AnotherKerasLayer(....)(out1)
....
model = Model(inp,out2)
#above is the usual process of creating a model
#supposing layer 2 is the layer you want (you can also use names)
weights = numpy.load(...path to your saved weights)
biases = numpy.load(... path to your saved biases)
model.layers[2].set_weights([weights,biases])
You can make layers untrainable (must be done before the model compilation):
model.layers[2].trainable = False
Then you compile the model:
model.compile(.....)
And there you go, a model, whose one layer is untrainable and has weights and biases defined by you, taken from somewhere else.
Yes, it is a common practice in transfer learning, see here.
Thjs piece_to_share below can be one or more layers.
piece_to_share = tf.keras.Model(...)
full_model = tf.keras.Sequential([piece_to_share, ...])
full_model.fit(...)
piece_to_share.save(...)