CODEWITHSUNDEEP
pythontensorflowkerastransfer-learning
markus-brln
markus-brln

Reputation: 25

Keras Transfer learning with own model

EDIT:

Apparently some issues with keras vs tensorflow.keras caused my problem. I followed this to change my imports: The added layer must be an instance of class Layer. Found: <tensorflow.python.keras.engine.input_layer.InputLayer>

I'm trying to load a model and exchange the last dense layer with another one of different output dimension. This is what I've got so far:

The model as it is saved:

def create_model(n_timesteps=828 , n_features=1, n_outputs=7):
    dtype='float32'
    model = Sequential([
        Convolution1D(input_shape=(n_timesteps, n_features), kernel_size=3, filters=128, activation='relu', kernel_regularizer='l2', dtype=dtype, name="conv1"),
        MaxPool1D(pool_size=4, strides=4), # reduce size of the input by 4
        Dropout(0.3, dtype=dtype),
        Flatten(),
        Dense(32, activation='relu',  kernel_regularizer=tf.keras.regularizers.l2(), dtype=dtype, name="dense1"),
        Dropout(0.2, dtype=dtype),
        Dense(n_outputs, activation='softmax', dtype=dtype, name="dense2")
    ])

    return model

and the attempt to retrain:

def retrain(model_name="cnn_general"):
    model = tf.keras.models.load_model('saved_models\\' + model_name)                        
    model.trainable = False
    model.compile(optimizer='adam', loss='categorical_crossentropy', 
              metrics=['accuracy'])
    print(model.summary())

    output = Dense(1, activation='sigmoid')(model.layers[-1].output)

    retrained_model = Model(inputs=model.inputs, outputs=output)
    print(retrained_model.summary())

I get the error output:

Traceback (most recent call last):
  File "C:/.../CNN_heatbeats.py", line 233, in <module>
    retrain((X_train, y_train, X_test,  y_test, X_val, y_val))
  File "C:/.../CNN_heatbeats.py", line 161, in retrain
    output = Dense(10, activation='sigmoid')(model.layers[-1].output)
  File "C:\Miniconda\envs\tensorflow-gpu-cuda10\lib\site-packages\keras\backend\tensorflow_backend.py", line 75, in symbolic_fn_wrapper
    return func(*args, **kwargs)
  File "C:\Miniconda\envs\tensorflow-gpu-cuda10\lib\site-packages\keras\engine\base_layer.py", line 475, in __call__
    previous_mask = _collect_previous_mask(inputs)
  File "C:\Miniconda\envs\tensorflow-gpu-cuda10\lib\site-packages\keras\engine\base_layer.py", line 1441, in _collect_previous_mask
    mask = node.output_masks[tensor_index]
AttributeError: 'Node' object has no attribute 'output_masks'

I've only found information on models like VGG16 but cannot apply that to my own network. Is what I've done in the retrain() function an appropriate approach? How can I make it work?

Upvotes: 0

Views: 464

Answers (1)

SELLAM
SELLAM

Reputation: 71

have tried it this way?

def retrain(model_name="cnn_general"):
    model = tf.keras.models.load_model('saved_models\\' + model_name)                        
    model.trainable = False
    model.compile(optimizer='adam', loss='categorical_crossentropy', 
              metrics=['accuracy'])
    model.summary()

    retrained_model = Sequential()
    for layer in model.layers[:-1]:
        layer.trainable = False
        retrained_model.add(layer)
    retrained_model.add(Dense(1, activation='sigmoid'))
    retrained_model.summary()

Upvotes: 1

Related Questions