Argthtjtr

I need to process some layers in a different way, doing some OR operations. I've found how to do it, I create a Lambda Layer and process the data with keras.backend.any. I am also doing a split, because I need to operate 2 separates groups with my logical OR.

def logical_or_layer(x):

    """Processing an OR operation"""

    import keras.backend

    #normalized to 0,1 

    aux_array = keras.backend.sign(x)

    aux_array = keras.backend.relu(aux_array)

    # OR operation

    aux_array = keras.backend.any(aux_array)

    # casting back the True/False to 1,0

    aux_array = keras.backend.cast(aux_array, dtype='float32')



    return aux_array

Then I'am creating my layers like this:

#this is the input tensor

inputs = Input(shape=(inputSize,))



#this is the Neurule layer

x = Dense(neurulesQt, activation='softsign')(inputs)

#after each neurule layer, the outputs need to be put into SIGNUM (-1 or 1)

x = Lambda(signumTransform, output_shape=lambda x:x, name='signumAfterNeurules')(x)



#separating into 2 (2 possible outputs)

layer_split0 = Lambda( lambda x: x[:, :end_output0], output_shape=(11, ), name='layer_split0')(x)

layer_split1 = Lambda( lambda x: x[:, start_output1:end_output1], output_shape=(9,), name='layer_split1')(x)



#this is the OR layer

y_0 = Lambda(logical_or_layer, output_shape=(1,), name='or0')(layer_split0)

y_1 = Lambda(logical_or_layer, output_shape=(1,), name='or1')(layer_split1)

Just FYI: Neurules are Neurons created based on IF-THEN rules, that's one project to work with Neurons which were trained with a TruthTable, representing Expert Knowledge.

Now, when I try to put the splitted layers back like this:

y = concatenate([y_0,y_1])

This error comes:

ValueError: Can't concatenate scalars (use tf.stack instead) for 'concatenate_32/concat' (op: 'ConcatV2') with input shapes: , , .

Then ok, let's use the tf.stack as suggested:

y = keras.backend.stack([y_0, y_1])

Then it can't be used as an output in the Model anymore, when I try:

model = Model(inputs=inputs, outputs=y)

Comes the error:

ValueError: Output tensors to a Model must be the output of a Keras `Layer` (thus holding past layer metadata). Found: Tensor("stack_14:0", shape=(2,), dtype=float32)

Checking with the function keras.backend.is_keras_tensor(y) it gives me False, but with all the other layers it gives me True

How should I concatenate it correctly?

EDIT: Following the answer from @today I was able to create a new Lambda Layer with the stack wrapped inside of it. But the outputs are modified, it should be (None,2) and it is (2,None,1) here is the output from model.summary():

__________________________________________________________________________________________________

Layer (type)                    Output Shape         Param #     Connected to                     

==================================================================================================

input_90 (InputLayer)           (None, 24)           0                                            

__________________________________________________________________________________________________

dense_90 (Dense)                (None, 20)           500         input_90[0][0]                   

__________________________________________________________________________________________________

signumAfterNeurules (Lambda)    (None, 20)           0           dense_90[0][0]                   

__________________________________________________________________________________________________

layer_split0 (Lambda)           (None, 11)           0           signumAfterNeurules[0][0]        

__________________________________________________________________________________________________

layer_split1 (Lambda)           (None, 9)            0           signumAfterNeurules[0][0]        

__________________________________________________________________________________________________

or0 (Lambda)                    (None, 1)            0           layer_split0[0][0]               

__________________________________________________________________________________________________

or1 (Lambda)                    (None, 1)            0           layer_split1[0][0]               

__________________________________________________________________________________________________

output (Lambda)                 (2, None, 1)         0           or0[0][0]                        

                                                                 or1[0][0]                        

==================================================================================================

Total params: 500

Trainable params: 0

Non-trainable params: 500

__________________________________________________________________________________________________

How should I define the output_shape in the layers to have the batch still there at the end?

EDIT2: Following the tips from @today I've done the following:

#this is the input tensor

inputs = Input(shape=(inputSize,))



#this is the Neurule layer

x = Dense(neurulesQt, activation='softsign')(inputs)

#after each neuron layer, the outputs need to be put into SIGNUM (-1 or 1)

x = Lambda(signumTransform, output_shape=lambda x:x, name='signumAfterNeurules')(x)

#separating into 2 (2 possible outputs)

layer_split0 = Lambda( lambda x: x[:, :end_output0], output_shape=[11], name='layer_split0')(x)    

layer_split1 = Lambda( lambda x: x[:, start_output1:end_output1], output_shape=[9], name='layer_split1')(x)

#this is the OR layer

y_0 = Lambda(logical_or_layer, output_shape=(1,), name='or0')(layer_split0)

y_1 = Lambda(logical_or_layer, output_shape=(1,), name='or1')(layer_split1)



y = Lambda(lambda x: K.stack([x[0], x[1]]),output_shape=(2,),  name="output")([y_0, y_1])

Now it seems to work correctly, the model.summary() below:

__________________________________________________________________________________________________

Layer (type)                    Output Shape         Param #     Connected to                     

==================================================================================================

input_1 (InputLayer)            (None, 24)           0                                            

__________________________________________________________________________________________________

dense_1 (Dense)                 (None, 20)           500         input_1[0][0]                    

__________________________________________________________________________________________________

signumAfterNeurules (Lambda)    (None, 20)           0           dense_1[0][0]                    

__________________________________________________________________________________________________

layer_split0 (Lambda)           (None, 11)           0           signumAfterNeurules[0][0]        

__________________________________________________________________________________________________

layer_split1 (Lambda)           (None, 9)            0           signumAfterNeurules[0][0]        

__________________________________________________________________________________________________

or0 (Lambda)                    (None, 1)            0           layer_split0[0][0]               

__________________________________________________________________________________________________

or1 (Lambda)                    (None, 1)            0           layer_split1[0][0]               

__________________________________________________________________________________________________

output (Lambda)                 (None, 2)            0           or0[0][0]                        

                                                                 or1[0][0]                        

==================================================================================================

Total params: 500

Trainable params: 0

Non-trainable params: 500

__________________________________________________________________________________________________

Wrap the K.stack inside a Lambda layer like this:

from keras import backend as K



y = Lambda(lambda x: K.stack([x[0], x[1]]))([y_0, y_1])