When to use 'use_bias=False' in Conv Nets (like in the Xception implementation)

[Andrés M.]

Looking at the Xception CNN example included in Keras I noticed that Chollet uses 'use_bias=False' in all the Convolutional layers.  

I was wondering, when to use it? or why?  can someone explain it to me or point me in the right direction? 

Thanks

[MohammadAli Dastgheib]

I have the same question. please help.

[Philip May]

When using batch normalization you can turn off bias.

See the paper here: https://arxiv.org/abs/1502.03167

"Note that, since we normalizeWu+b, the biasbcan beignored since its effect will be canceled by the subsequentmean subtraction [..]"