Adding a custom activation function in Keras

Question

I am trying to implement a custom activation function in Keras The function is following

def picewise(x):
   if x > 0.5 :
      return 1
   elif if x < 0.5:
      return 0
   else:
      return x + 0.5

I call the activation function as

model.add(Dense(128,activation = picewise))

But I am getting a whole bunch of errors. How can i successfully implement the above as a working activation function in Keras

Md Johirul Islam · Accepted Answer · 2018-05-02 05:06:33Z

3

Here x is not a primitive type. x is a tensor. So the way you are trying to implement the activation function will not work. You probably need something like the following:

import tensorflow as tf
def picewise(x):
    z = tf.where(x >= 0.5, x - x + 1.0 , x)
    y = tf.where(z <= -0.5,z -  z + 0, z + 0.5)
    return y

answered May 2, 2018 at 5:06

Md Johirul Islam

5,1924 gold badges28 silver badges57 bronze badges

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

Collectives™ on Stack Overflow

Adding a custom activation function in Keras

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related