Keras Functional API Multi Input Layer

Question

How do I define a multi input layer using Keras Functional API? Below is an example of the neural network I want to build. There are three input nodes. I want each node to be a 1 dimensional numpy array of different lengths.

Here's what I have so far. Basically I want to define an input layer with multiple input tensors.

from keras.layers import Input, Dense, Dropout, concatenate
from keras.models import Model

x1 = Input(shape =(10,))
x2 = Input(shape =(12,))
x3 = Input(shape =(15,))

input_layer = concatenate([x1,x2,x3])

hidden_layer = Dense(units=4, activation='relu')(input_layer)
prediction = Dense(1, activation='linear')(hidden_layer)

model = Model(inputs=input_layer,outputs=prediction)

model.summary()

The code gives the error.

ValueError: Graph disconnected: cannot obtain value for tensor Tensor("x1_1:0", shape=(?, 10), dtype=float32) at layer "x1". The following previous layers were accessed without issue: []

Later when I fit the model I will pass in a list of 1D numpy arrays with the corresponding lengths.

The funcional API guide describes exactly how to make multiple input models keras.io/getting-started/functional-api-guide your problem is that you have to give the inputs (x1, x2, x3), not the layer after that. — Dr. Snoopy
– Dr. Snoopy, Commented Jan 3, 2019 at 21:16

Primusa · Accepted Answer · 2019-01-03 21:18:02Z

3

The inputs must be your Input() layers:

model = Model(inputs=[x1, x2, x3],outputs=prediction)

answered Jan 3, 2019 at 21:18

Primusa

13.5k3 gold badges36 silver badges56 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

Luke DeLuccia · Accepted Answer · 2019-01-03 21:19:19Z

2

Change

model = Model(inputs=input_layer,outputs=prediction)

to

model = Model(inputs=[x1, x2, x3],outputs=prediction)

answered Jan 3, 2019 at 21:19

Luke DeLuccia

5416 silver badges16 bronze badges

Collectives™ on Stack Overflow

Keras Functional API Multi Input Layer

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related