search

immortal3 / AutoEncoder-Based-Communication-System

Tensorflow Implementation and result of Auto-encoder Based Communication System From Research Paper : "An Introduction to Deep Learning for the Physical Layer" http://ieeexplore.ieee.org/document/8054694/
126 stars 79 forks source link

batch normalization layer need to be corrected #3

Closed safarisadegh closed 4 years ago

safarisadegh commented 5 years ago

hi, i think you made a mistake in using keras BatchNormalization in autoencoder_7_4_symbol.ipynb. If you see in model summary you see that batch normalization layer has 14 trainable parameters that is because of scale and center parameters have True default value. I think you should use BatchNormalization as follows: encoded2 = BatchNormalization(center=False, scale=False)(encoded1)

Thank you for your helpfull repo.

summury
immortal3 commented 4 years ago

Yes, It looks like I implemented it wrongly. But, You can refer to autoencoder_dynamic.ipynb for correct results. Closing this.