CODEWITHSUNDEEP
pythontensorflowkerastf.keras
Gary
Gary

Reputation: 863

How to set the default parameters of Conv2D in tf.keras?

Support i have a network with 5 convolution. I write it by Keras.

x = Input(shape=(None, None, 3))
y = Conv2D(10, 3, strides=1)(x)
y = Conv2D(16, 3, strides=1)(y)
y = Conv2D(32, 3, strides=1)(y)
y = Conv2D(48, 3, strides=1)(y)
y = Conv2D(64, 3, strides=1)(y)

I want set all convolution's kernel_initializer to xavier. One of method is: 

x = Input(shape=(None, None, 3))
y = Conv2D(10, 3, strides=1, kernel_initializer=tf.glorot_uniform_initializer())(x)
y = Conv2D(16, 3, strides=1, kernel_initializer=tf.glorot_uniform_initializer())(y)
y = Conv2D(32, 3, strides=1, kernel_initializer=tf.glorot_uniform_initializer())(y)
y = Conv2D(48, 3, strides=1, kernel_initializer=tf.glorot_uniform_initializer())(y)
y = Conv2D(64, 3, strides=1, kernel_initializer=tf.glorot_uniform_initializer())(y)

But this kind of writing is very sad and the code is very redundant.

Is there a better way of writing?

Upvotes: 0

Views: 561

Answers (2)

Sreeram TP
Sreeram TP

Reputation: 11917

Better make a lambda that will make a Conv2D layer and fix the initializer as needed and call it in the model definition part.

I think lambda is more suitable in this situation than a function.

You can do it like this,

customConv = lambda filters, kernel : Conv2D(filters, kernel, strides=1, kernel_initializer=tf.glorot_uniform_initializer())

x = Input(shape=(None, None, 3))

y = customConv(10, 3)(x)
y = customConv(16, 3)(y)
y = customConv(32, 3)(y)
y = customConv(48, 3)(y)
y = customConv(64, 3)(y)

Upvotes: 1

Dr. Snoopy
Dr. Snoopy

Reputation: 56367

Keras provides no way to change the defaults, so you can just make a wrapper function:

def myConv2D(filters, kernel):
    return Conv2D(filters, kernel, strides=1, kernel_initializer=tf.glorot_uniform_initializer())

And then use it as:

x = Input(shape=(None, None, 3))
y = myConv2D(10, 3)(x)
y = myConv2D(16, 3)(y)
y = myConv2D(32, 3)(y)
y = myConv2D(48, 3)(y)
y = myConv2D(64, 3)(y)

Upvotes: 3

Related Questions