I have the following code. When I am trying to merge([user_latent, item_latent], mode = 'concat') I am getting
Shape (2, ?, ?) and () are incompatible. On searching, it says that it is tensorflow API issue but I am not sure how can I change it since, I am not using tf.concat().
def get_model(num_users, num_items, layers = [20,10], reg_layers=[0,0]):
assert len(layers) == len(reg_layers)
num_layer = len(layers) #Number of layers in the MLP
# Input variables
user_input = Input(shape=(1,), dtype='int32', name = 'user_input')
item_input = Input(shape=(1,), dtype='int32', name = 'item_input')
MLP_Embedding_User = Embedding(input_dim = num_users, output_dim = layers[0]/2, name = 'user_embedding',
init = init_normal, W_regularizer = l2(reg_layers[0]), input_length=1)
MLP_Embedding_Item = Embedding(input_dim = num_items, output_dim = layers[0]/2, name = 'item_embedding',
init = init_normal, W_regularizer = l2(reg_layers[0]), input_length=1)
# Crucial to flatten an embedding vector!
user_latent = Flatten()(MLP_Embedding_User(user_input))
item_latent = Flatten()(MLP_Embedding_Item(item_input))
# The 0-th layer is the concatenation of embedding layers
vector = merge([user_latent, item_latent], mode = 'concat')
# MLP layers
for idx in xrange(1, num_layer):
layer = Dense(layers[idx], W_regularizer= l2(reg_layers[idx]), activation='relu', name = 'layer%d' %idx)
vector = layer(vector)
# Final prediction layer
prediction = Dense(1, activation='sigmoid', init='lecun_uniform', name = 'prediction')(vector)
model = Model(input=[user_input, item_input],
output=prediction)
return model
Related
I have various inputs, some that need embedding. I have been able to create them all as seen below:
I can then concatenate them all, for the following:
However, my disconnect is where to go from here. I have built the following autoencoder, but I am not sure how to "stack" the previous embedding+input mix on top of this flow:
So, how do I make the input layer whats already been defined above? I tried setting the first "encoder" part to take in merge_models, but it fails:
Code is the following:
num_input = Input(shape=scaled_data.shape[1], name='input_number_features')
models.append(num_input)
inputs.append(num_input)
binary_input = Input(shape=binary_data.shape[1], name='input_binary_features')
models.append(binary_input)
inputs.append(binary_input)
for var in cols_to_embed :
model = Sequential()
no_of_unique_cat = data[var].nunique()
embedding_size = np.ceil(np.sqrt(no_of_unique_cat))
embedding_size = int(embedding_size)
print(var + " - " + str(no_of_unique_cat) + ' unique values to ' + str(embedding_size))
inpt = tf.keras.layers.Input(shape=(1,),\
name='input_' + '_'.join(\
var.split(' ')))
embed = tf.keras.layers.Embedding(no_of_unique_cat, embedding_size,trainable=True,\
embeddings_initializer=tf.initializers\
.random_normal)(inpt)
embed_rehsaped = tf.keras.layers.Reshape(target_shape=(embedding_size,))(embed)
models.append(embed_rehsaped)
inputs.append(inpt)
merge_models = tf.keras.layers.concatenate(models)
# Input Layer
input_dim = merge_models.shape[1]
input_layer = Input(shape = (input_dim, ), name = 'input_layer')
# Encoder
encoder = Dense(16, activation='relu')(input_layer)
encoder = Dense(8, activation='relu')(encoder)
encoder = Dense(4, activation='relu')(encoder)
# Bottleneck
z = Dense(2, activation='relu')(encoder)
# Decoder
decoder = Dense(4, activation='relu')(z)
decoder = Dense(8, activation='relu')(decoder)
decoder = Dense(16, activation='relu')(decoder)
decoder = Dense(input_dim, activation='elu')(decoder) # intentionally using 'elu' instead of 'reul'
# Autoencoder
from tensorflow.keras.models import Model
autoencoder = Model(inputs = input_layer,
outputs = decoder,
name = 'ae_toy_example')
You should pass merge_models into the first encoder layer in this way:
encoder = Dense(16, activation='relu')(merge_models)
then you should define your final model in this way:
Model(inputs = inputs, outputs = decoder, name = 'ae_toy_example')
and NOT as:
Model(inputs = input_layer, outputs = decoder, name = 'ae_toy_example')
I would like to know why for two same keras models, sometimes get_method() gives the same results (See model_dense_A and model_dense_B) and sometimes not (example for model_conv_A and model_conv_B).
Even if I use the clear_session() method and the exact same code, the models are still different.
Does someone know about this behaviour ?
Code snippet:
from tensorflow import keras
input_shape = (300, 3)
# MODEL DENSE A
keras.backend.clear_session()
input_ = keras.layers.Input(shape=input_shape)
out = keras.layers.Flatten()(input_)
out = keras.layers.Dense(units=100, activation='relu')(out)
out = keras.layers.Dense(units=50, activation='relu')(out)
out = keras.layers.Dense(units=10, activation='softmax')(out)
mdl_dense_A = keras.models.Model(name='dense', inputs=input_, outputs=out)
# MODEL DENSE B
keras.backend.clear_session()
input_ = keras.layers.Input(shape=input_shape)
out = keras.layers.Flatten()(input_)
out = keras.layers.Dense(units=100, activation='relu')(out)
out = keras.layers.Dense(units=50, activation='relu')(out)
out = keras.layers.Dense(units=10, activation='softmax')(out)
mdl_dense_B = keras.models.Model(name='dense', inputs=input_, outputs=out)
print(mdl_dense_A.get_config() == mdl_dense_B.get_config()) # True
# MODEL CONV1D-LSTM A
keras.backend.clear_session()
input_ = keras.layers.Input(shape=input_shape)
branch_outputs = []
for i in range(input_shape[-1]):
out = keras.layers.Lambda(lambda x: keras.backend.expand_dims(x[:, :, i], axis=-1))(input_)
out = keras.layers.Conv1D(filters=20, kernel_size=5, strides=2, padding='valid')(out)
branch_outputs.append(out)
out = keras.layers.Concatenate()(branch_outputs)
recursive = keras.layers.LSTM(50, return_sequences=True)(out)
recursive = keras.layers.LSTM(50)(recursive)
out = keras.layers.Dense(10, activation='softmax')(recursive)
mdl_conv_A = keras.models.Model(name='conv1d-lstm', inputs=input_, outputs=out)
# MODEL CONV1D-LSTM B
keras.backend.clear_session()
input_ = keras.layers.Input(shape=input_shape)
branch_outputs = []
for i in range(input_shape[-1]):
out = keras.layers.Lambda(lambda x: keras.backend.expand_dims(x[:, :, i], axis=-1))(input_)
out = keras.layers.Conv1D(filters=20, kernel_size=5, strides=2, padding='valid')(out)
branch_outputs.append(out)
out = keras.layers.Concatenate()(branch_outputs)
recursive = keras.layers.LSTM(50, return_sequences=True)(out)
recursive = keras.layers.LSTM(50)(recursive)
out = keras.layers.Dense(10, activation='softmax')(recursive)
mdl_conv_B = keras.models.Model(name='conv1d-lstm', inputs=input_, outputs=out)
print(mdl_conv_A.get_config() == mdl_conv_B.get_config()) # False !?
It usually helps to take a look at where exactly both configs differ, i.e.
print(mdl_conv_A.get_config() == mdl_conv_B.get_config(), (mdl_conv_A.get_config(), mdl_conv_B.get_config())) # False !?
In this case they differ because of the lambda layer, which is not very serializable.
I am building a small network using some custom network boxes for each use case, It looks like this :
def top_block(dropout = None, training = None):
# scaled input
input_1 = tf.keras.Input(shape=(1,15), dtype='float32')
input_2 = tf.keras.Input(shape=(1,15), dtype='float32')
if dropout:
layer_one = tf.keras.layers.Dropout(rate = dropout)(input_1, training = training)
layer_two = tf.keras.layers.Dropout(rate = dropout)(input_2, training = training)
return [layer_one,layer_two]
return [input_1,input_2]
def bottom_layer(input_layers):
data = tf.reduce_mean(input_layers,0)
cls_layer = tf.keras.layers.Dense(1,
kernel_initializer = keras.initializers.glorot_uniform(seed=200),
activation = 'sigmoid')(data)
model = tf.keras.Model([input_layers[0], input_layers[1]], cls_layer , name = 'model_1')
model.compile(loss = 'binary_crossentropy', optimizer = 'adam', metrics=['accuracy'])
model.summary()
return model
If I am trying to access this network without dropout, it's working fine :
top_ = top_block()
model = bottom_layer(top_ )
But if I am accessing with dropout, it's giving error:
top_ = top_block(dropout = 0.2, training = True)
model = bottom_layer(top_ )
ValueError: Graph disconnected: cannot obtain value for tensor Tensor("input_72:0", shape=(None, 1, 15), dtype=float32) at layer "input_72". The following previous layers were accessed without issue: []
How to access the model with dropout layer?
How to disable training = False during evaluate? Do I need to load full model and old model weights?
Thank You!
I just realized my input is coming from intermediate layer (dropout layer), It should come directly from Input layer :
def top_block():
# scaled input
input_1 = tf.keras.Input(shape=(1,15), dtype='float32')
input_2 = tf.keras.Input(shape=(1,15), dtype='float32')
return [input_1, input_2]
def apply_dropout(layers_data, dropout_val, training):
layer_one = tf.keras.layers.Dropout(rate = dropout_val)(layers_data[0], training = training)
layer_two = tf.keras.layers.Dropout(rate = dropout_val)(layers_data[1], training = training)
return [layer_one, layer_two]
def bottom_layer(input_layers, data):
data = tf.reduce_mean(data, 0)
cls_layer = tf.keras.layers.Dense(1,
kernel_initializer = keras.initializers.glorot_uniform(seed=200),
activation = 'sigmoid')(data)
model = tf.keras.Model(input_layers, cls_layer , name = 'model_1')
model.compile(loss = 'binary_crossentropy', optimizer = 'adam', metrics=['accuracy'])
model.summary()
return model
It's working now
top_ = top_block()
dropout_ = apply_dropout(top_, 0.2, True)
model = bottom_layer(top_ , dropout_)
How can i use saved weights of the two identical nets in a Siamese Network -> similarity net (right/ siamese) and the feature generator (left).
I have followed and implemented this example for image similarity and saved the weights of two identical nets in the siames model as simlarity_model.h5 and feature_generator_model.h5.
How can I use these two saved weights to do predictions?
Similarity model =>
from keras.layers import concatenate
img_a_in = Input(shape = x_train.shape[1:], name = 'ImageA_Input')
img_b_in = Input(shape = x_train.shape[1:], name = 'ImageB_Input')
img_a_feat = feature_model(img_a_in)
img_b_feat = feature_model(img_b_in)
combined_features = concatenate([img_a_feat, img_b_feat], name = 'merge_features')
combined_features = Dense(16, activation = 'linear')(combined_features)
combined_features = BatchNormalization()(combined_features)
combined_features = Activation('relu')(combined_features)
combined_features = Dense(4, activation = 'linear')(combined_features)
combined_features = BatchNormalization()(combined_features)
combined_features = Activation('relu')(combined_features)
combined_features = Dense(1, activation = 'sigmoid')(combined_features)
similarity_model = Model(inputs = [img_a_in, img_b_in], outputs = [combined_features], name = 'Similarity_Model')
similarity_model.summary()
Feature generator model =>
from keras.models import Model
from keras.layers import Input, Conv2D, BatchNormalization, MaxPool2D, Activation, Flatten, Dense, Dropout
img_in = Input(shape = x_train.shape[1:], name = 'FeatureNet_ImageInput')
n_layer = img_in
for i in range(2):
n_layer = Conv2D(8*2**i, kernel_size = (3,3), activation = 'linear')(n_layer)
n_layer = BatchNormalization()(n_layer)
n_layer = Activation('relu')(n_layer)
n_layer = Conv2D(16*2**i, kernel_size = (3,3), activation = 'linear')(n_layer)
n_layer = BatchNormalization()(n_layer)
n_layer = Activation('relu')(n_layer)
n_layer = MaxPool2D((2,2))(n_layer)
n_layer = Flatten()(n_layer)
n_layer = Dense(32, activation = 'linear')(n_layer)
n_layer = Dropout(0.5)(n_layer)
n_layer = BatchNormalization()(n_layer)
n_layer = Activation('relu')(n_layer)
feature_model = Model(inputs = [img_in], outputs = [n_layer], name = 'FeatureGenerationModel')
feature_model.summary()
Saved weights =>
feature_model.save('feature_model.h5')
similarity.save('feature_model.h5')
I am trying to build a model that classifies images from belgium traffic datasets, but spyder freezes when i try to run the optimizer and train operation in sess.run[].
i have attached the code below.
graph = tf.Graph()
with graph.as_default():
images_X = tf.compat.v1.placeholder(tf.float32,shape = [None,32,32,3])
labels_X = tf.compat.v1.placeholder(tf.int32,shape = [None])
#biasInit = tf.constant_initializer(0.1, dtype=tf.float32)
# Initializer
biasInit = tf.initializers.GlorotUniform()
#conv layer 1 - num_filtewrs = 128, kernel size = [6,6]
conv_1 = Conv2D(filters = 128,kernel_size = [6,6],bias_initializer = biasInit)(images_X)
#batch normalization
bn_1 = BatchNormalization(center = True,scale = True)(conv_1)
#maxpoloing
pool_1 = MaxPooling2D(pool_size = (2,2))(bn_1)
#conv layer 2
conv_2 = Conv2D(filters = 256,kernel_size = [6,6] , strides = (2,2),
bias_initializer=biasInit)(pool_1)
#Batch normalization 2
bn_2 = BatchNormalization(center = True,scale = True)(conv_2)
pool_2 = MaxPooling2D(pool_size = (2,2))(bn_2)
#faltten
images_flat = Flatten()(pool_2)
#dense layer - units = 512
fc_1 = Dense(units = 512,activation = 'relu')(images_flat)
bn_3 = BatchNormalization(center = True,scale = True)(fc_1)
dropout = Dropout(0.25)(bn_3)
#logits will be of the size [None,62]
logits = Dense(units = 62,activation = 'relu')(dropout)
#converting the lofits - [None,62] to labels - [None]
predicted_labels = tf.argmax(logits, axis=1)
loss_op = tf.reduce_mean(tf.nn.sparse_softmax_cross_entropy_with_logits(logits = logits,
labels = labels_X))
update_ops = tf.compat.v1.get_collection(tf.compat.v1.GraphKeys.UPDATE_OPS)
with tf.control_dependencies(update_ops):
# Create an optimizer, which acts as the training op.
train = tf.compat.v1.train.AdamOptimizer(learning_rate=0.10).minimize(loss_op)
init_op = tf.compat.v1.global_variables_initializer()
print("images_flat: ", images_flat)
print("logits: ", logits)
print("loss: ", loss_op)
print("predicted_labels: ", predicted_labels)
# Create a session to run the graph we created.
session = tf.compat.v1.Session(graph=graph, config=tf.compat.v1.ConfigProto(log_device_placement=True))
session.run(init_op)
for i in range(10):
_, loss_value = session.run([train, loss_op], feed_dict={images_X: images_array, labels_X: labels_array})
print("Loss: ", loss_value)
when i run the program line by line, it runs well until the last for loop, when it reaches the last for loop the memory goes to 99% and the whole pc freezes.